Murdoch Dissertation (c)2007

A FUNCTIONAL GROUP APPROACH FOR PREDICTING THE COMPOSITION OF HARD CORAL ASSEMBLAGES IN FLORIDA AND BERMUDA

A Dissertation Submitted to the Graduate Faculty of the University of South Alabama in partial fulfillment of the requirements for the degree of Doctor of Philosophy in The Department of Marine Science

by Thaddeus J. T. Murdoch B.Sc., Dalhousie University, 1988 B.A. Honours, Dalhousie University, 1991 M.S., University of South Alabama, 1998 Copyright c 2007 Thaddeus J. T. Murdoch All rights reserved

i

ii

ACKNOWLEDGEMENTS

The Keyswide Coral Reef Expedition was funded by NOAA’s National Undersea Research Center at the University of North Carolina, Wilmington; NOAA’s Sanctuaries and Reserves Division; The U.S. National Park Service, Biscayne National Park; The Munson Foundation; The Florida Institute of Oceanography; and the Harbour Branch Oceanographic Institution. The Expedition was carried out under permits from the Florida Keys National Marine Sanctuary, the U.S. National Park Service, and the State of Florida. The Bermuda Project was supported by a fellowship from the University of South Alabama; a PADI Aware grant; and a Bermuda Programme award from the Bermuda Institute of Ocean Sciences. Additional support was provided by the Bermuda Biodiversity Project, Bermuda Zoological Society, the Ernest E. Stempel Foundation, and the Department of Conservation Services, Bermuda Government. The research in this dissertation could not have been done without the help of many people. I am grateful to John Ogden and Steven Miller for organizing the Keyswide Coral Reef Expedition, and to Ken Johns and Otto Rutten for running the program of continuous nitrox diving during the cruises. They, Dennis Hanisak and Laura Seimon participated in the field work that formed the basis of the Florida section of this dissertation. Dione Swanson provided valuable assistance in both the field and laboratory. I am grateful to my committee members, and the professors and students of the Dauphin

iii

Island Sea Lab and the Bermuda Institute of Ocean Sciences for their assistance, interest and support of this project. Field work in Bermuda would not have been possible without the lively assistance of the members of Robbie Smith’s Benthic Ecology Research Lab during 2000 – 2003 and the BREAM team at BZS from 2004 – 2006. I am also thankful to Annie Glasspool and Jack Ward for supporting me while I wrote up the dissertation. Jon Martin, Julie Prerost, Toby Bolton, Jeannette Loram, Alexander Venn , Philippe Rouja, Mike Colella, Gerardo Toro Farmer, and Matt Ajemian provided invaluable scientific and moral support. I am deeply indebted to my family for unwavering encouragement.

iv

TABLE OF CONTENTS

LIST OF TABLES..............................................................................................................ix LIST OF FIGURES...........................................................................................................xii

ABSTRACT.......................................................................................................................xx

CHAPTER 1: THE NEED FOR A FUNCTIONAL GROUP APPROACH TO DESCRIBE THE STRUCTURE OF CARIBBEAN HARD CORAL ASSEMBLAGES.........................................................................................1

Introduction..................................................................................................1 Functional traits and functional groups in reef corals..................................4 Assigning species to functional groups using the Adaptive Strategies Theory..........................................................................................6 Characteristics of each Adaptive Strategy..................................................12 The graphic model of the Adaptive Strategies Theory...............................15 Assigning Caribbean reef corals to functional groups and defining the critical tests............................................................................21

v

Characteristics of each functional group of coral.......................................27 Sources of environmental stress and disturbance on coral reefs................32 Testing the applicability of the functional group approach for reef corals..............................................................................................35 Similarities in distribution of species between and among FG......35 Rank abundance by species and functional groups........................39 Percent cover of all corals..............................................................40 Percent cover per functional group................................................41 Total species richness.....................................................................45 Functional group richness..............................................................47 Species richness within functional groups.....................................52 Testing the adaptive strategies theory on Caribbean coral reefs................54

CHAPTER 2: THE RESPONSES OF FUNCTIONAL GROUPS OF CORALS TO DIRECT AND INDIRECT GRADIENTS ON THE FLORIDA REEF TRACT.......................................................................................................56

Introduction................................................................................................56 Objectives...................................................................................................61 Similarities in distribution of species between and among FG......62 Rank abundance by species and functional groups........................63

vi

Percent cover and abundance per functional group.......................64 Species richness of all corals.........................................................66 Functional group richness..............................................................66 Species richness per functional group............................................67

Methodology...............................................................................................68 Geographic setting.........................................................................68 Data collection and analysis...........................................................69 Statistical Analysis.........................................................................73

Results ........................................................................................................75 Similarities in distribution of species between and among FG......75 Rank abundance by species...........................................................82 Rank abundance per functional group...........................................88 Percent cover of each functional group vs. W................................92 Percent cover of each functional group vs. total coral cover.........94 Total species richness vs. W...........................................................98 Total species richness vs. total coral cover....................................99 Functional group richness vs. W..................................................101 Functional group richness vs. total coral cover...........................105 Species richness within functional groups vs. W.........................109 Species richness within functional groups vs. total coral cover. .112

vii

Discussion.................................................................................................118 Dominance by species and functional groups..............................118 Percent cover per functional group..............................................121 Total species richness...................................................................123 Functional group richness............................................................124 Species richness within functional groups...................................125 Conclusions..............................................................................................128

CHAPTER 3: THE GEOGRAPHY AND ENVIRONMENTAL CHARACTERISTICS OF THE NORTH LAGOON OF BERMUDA..................................................................................133

Introduction...................................................................................133 Previous research into the distribution of corals across the North Lagoon..........................................................................................141 Predominant environmental factors in operation across the study area

143 Suspended particulate matter...........................................143 Water temperature............................................................146 Solar radiation..................................................................147 The effect of depth and turbidity.........................147 The effect of aspect..............................................148

viii

Wave energy and currents................................................150

Project 1: Mapping of the lagoonal reefs.....................................151 Introduction and Methodology........................................151 Results and Discussion....................................................154 Project 2: Assessing the three dimensional light field over a range of depths............................................................................................. 157 Introduction and Methodology........................................156 Results and Discussion....................................................157 Project 3: Cross-platform differences in downwelling light availability....................................................................................162 Introduction and Methodology........................................162 Results and Discussion....................................................164 Discussion.....................................................................................169

CHAPTER 4: THE DISTRIBUTION OF CORAL SPECIES AND FUNCTIONAL GROUPS OVER PHYSICAL GRADIENTS ACROSS THE NORTH LAGOON OF BERMUDA......................................................................173

Introduction..............................................................................................173 Objectives.................................................................................................175

ix

Similarities in the distribution of species and functional groups.175 Similarities among sites in species assemblages.........................178 Percent cover and abundance per functional group.....................178 Species distributions across sites.................................................179 Species richness of all corals.......................................................180 Functional group richness............................................................180 Methodology.............................................................................................181 Data collection.............................................................................181 Data analysis................................................................................186 Statistical analysis........................................................................190 Results ......................................................................................................192 Depths per reefs...........................................................................192 Similarity in species distributions across sites on each reef........195 Similarity among sites in species assemblages............................199 Distribution patterns of coral species...........................................215 Frequency of occurrence........................................................215 Standard measures for coral reefs................................................242 Section 1: Tops of reefs only..................................................242 A. Average percent coral cover........................................243 B. Species richness...........................................................243 C. Functional group richness...........................................243 D. Percent cover of the branched viviparous FG.............244 E. Percent cover of the massive viviparous FG...............244

x

F. Percent cover of the massive oviparous FG.................245 Section 2: Tops and south sides of reefs................................249 A. Average percent coral cover........................................249 B. Species richness...........................................................249 C. Functional group richness...........................................250 D. Percent cover of the branched viviparous FG.............250 E. Percent cover of the massive viviparous FG...............252 F. Percent cover of the massive oviparous FG.................252 G. Percent cover of the folious and plating viviparous FG..............................................................253 Discussion.................................................................................................262 Species and functional group distribution across sites................262 Percent cover per functional group..............................................264 Management issues......................................................................265

CHAPTER 5: SUMMARY, CONCLUSIONS AND RECOMMENDATIONS..............269

REFERENCES................................................................................................................283

APPENDICES.................................................................................................................306 Appendix A Digital video image capture methodology...........................308 Appendix B: Applescript computer program to place dots on frames.....313 Appendix C: A list of coral species observed in Bermuda.......................317

xi

Appendix D: Bermuda climatology..........................................................318 Appendix E: Logistic regression of rank abundances; Florida data.........319

BIOGRAPHICAL SKETCH...........................................................................................325

xii

LIST OF TABLES

Table Page

1.01

The characteristic differences in biological attributes between competitive, stress-tolerant and ruderal plant species (modified from Grime 1979)...........................................................................................10

1.02

Life history characteristics of massive scleractinian corals as defined in Table 3 in Soong (1993).........................................................................23

1.03

The ranking of each proposed functional group in ten critical traits, and the adaptive strategy to which they most closely represent...................26

xiii

2.01

Results of one-way analysis of similarity of the Bray-Curtis similarities of the abundance data of the most abundant 11 species observed across 20 reef sites located on the Florida Reef Tract......................................79

2.02

Results of one-way analysis of similarity of the Bray-Curtis similarities of the abundance data of all species observed across 20 reef sites located on the Florida Reef Tract...........................................................81

2.03

A table of the number of occurrences with which each the 36 most abundant species ranked from 1 to 20 across all 200 transects..............85

2.04

A table of the observed number of times each functional group ranked from 1 to 4 across the 200 transects surveyed.......................................90

2.05

Results of two-way t-test of the linear correlation between coral cover of each functional group and W on 10 reef sites along the Florida Reef Tract, testing whether the slopes are zero..............................................94

2.06

Results of orthogonal contrasts on whether the linear regressions of percent coral cover for each functional group versus total coral cover at each reef site were significantly different from zero.............................97

xiv

2.07

Results of two-way t-tests on whether the second-order coefficients for polynomial regressions of FG cover versus total coral cover were significantly different from zero............................................................97

2.08

Results of orthogonal contrasts on the linear regressions of functional group richness for each level of constraint versus W...........................102

2.09

Results of orthogonal contrasts on the second-order coefficients for polynomial regressions of functional group richness for each level of constraint versus W..............................................................................102

2.10

Presence or absence matrices of the presence or absence of functional groups across the ten reefs of the environmental gradient W. The rules for inclusion of functional groups are as in Figure 2.18, above..........104

2.11

Results of orthogonal contrasts on the linear regressions of functional group richness for each level of constraint versus W...........................107

2.12

Results of orthogonal contrasts on the second-order coefficients for polynomial regressions of functional group richness for each level of constraint versus W..............................................................................107

xv

2.13

Results of analyses of variance of the linear regression of species richness versus W for each functional group.......................................110

2.14

Results of two-way t-tests on whether the second-order coefficients for polynomial regressions for species richness of each functional group versus W were significantly different from zero..................................110

2.15

Results of two-way t-tests of the linear regression of species richness for each functional group versus total coral cover for each functional group....................................................................................................114

2.16

Results of two-way t-tests on whether the second-order coefficients for polynomial regressions of FG species richness versus total coral assemblage cover were significantly different from zero....................114

2.17

Sorted matrices of species presence or absence for each functional group across 20 transects at each of the 20 sites of the Keyswide Coral Reef Expedition....................................................................................116

3.01

Characteristics of each zone of the survey area across the Bermuda Lagoon, and the patch reefs contained therein.....................................155

xvi

3.02

Two-way analysis of variance of the effects of depth and aspect on the proportion of surface light reaching a hemispherical sensor...............161

3.03

ANOVA table of the differences in light intensity across the five locations from the data illustrated in Figure 2.13................................168

3.04

Results of a Tukey’s post hoc analysis of the significance in the differences in the amount of luminance at 8-m depth over the hours of 11 am to 1 pm between the five locations across the reef platform.....168

4.01

Details of the 18 surveyed reefs surveyed across the North Lagoon...185

4.02

Results of an ANOSIM analysis of distinctness in clustering of each functional group...................................................................................198

4.03

ANOSIM table for the factor Aspect across all sites, including the results of pairwise post-hoc tests.........................................................203

4.04

Significance levels of separate ANOSIM tests comparing similarities between reef sites located on the south versus north sides of reefs in each zone and at different depths.........................................................204

xvii

4.05

ANOSIM table for the factor Depth, calculated from data at each depth from across all sites..............................................................................207

4.06

Significance levels of pair-wise tests comparing similarities between reef sites located on different depths...................................................207

4.07

ANOSIM table for the factor Zone across all sites..............................210

4.08

SIMPER analysis of the dominant species that differ between zones across the Bermuda Platform...............................................................211

4.09

Results of the 2-way ANOVAs of the six parameters across the 18 reef sites located on the tops of patch reefs located in three replicate “legs” across six zones located across the north lagoon in Bermuda.............248

xviii

LIST OF FIGURES

Figure Page

1.1 Percent cover for each of the 38 species recorded on the 20 reefs surveyed on the Keyswide Coral Reef Expedition...................................................................3 1.02

A modified diagram of Grime’s (1979) Adaptive Strategy Theory for classifying habitats according to levels of stress and disturbance.........11

1.03

The Adaptive Strategies Theory graphic model, also known as the CSR model, depicted as a ternary diagram....................................................16

1.04

Generalized model of community dominants by Steneck and Dethier (1994) that refines to Grime’s (1977) AST model (shaded in light gray) ...............................................................................................................18

xix

1.05

A modified version of Grime’s (1977) and Steneck and Dethier’s (1994) generalized two-dimensional model of FG dominance within habitat types.......................................................................................................20

1.06

Representation of how varying levels of the four most important environmental factors act to promote growth, or act as a stressor or disturbance agent to corals....................................................................34

1.07

An illustration of different hypothetical distributions of species and functional groups across an environmental gradient (A to D), and how they appear when graphed as (i) abundances, (ii) tabulated in a matrix of presence vs. absence, and (iii) graphed using a multivariate ordination technique, such as Multidimensional Scaling (MDS)..........37

1.08

An illustration of how total assemblage biomass is predicted to vary across habitat types characterized by different levels of resource availability and disturbance...................................................................41

1.09

Diagrams depicting the differing ways in which the abundances of competitive (C), stress-tolerant (S) and ruderal (R) functional groups of corals are predicted to vary across habitats located across the range of stress and disturbance gradients encompassed by the AST mode.........43

xx

1.10

The distribution of species richness predicted to occur across habitats by Grime’s (1977) Adaptive Strategies model...........................................46

1.11

A diagram illustrating the range of strategies encompasses by (a) annual herbs, (b) biennial herbs, (c) perennial herbs and ferns, (d) trees and shrubs, (e) lichens and (f) bryophytes...................................................48

1.12

A three dimensional model of the levels of biomass predicted for set of functional groups of species across habitat types characterized by varying levels of disturbance potential and productivity potential........49

1.13

A diagram illustrating how functional groups are predicted to be dispersed across patches located with habitats defined by varying rates of resource gain and loss.......................................................................52

2.01

The elongated oval within this square diagram of state space represents the hypothetical range within the AST (CSR) model that was occupied by the sites of the Florida Reef Tract that are the focus of this chapter.........58

2.02

Relationship between total percent cover of the entire coral assemblage and the measure of environmental disturbance due to island passes, W ...............................................................................................................60

xxi

2.03

Map of South Florida and the Florida Keys...........................................69

2.04

A dendrogram showing the similarities in response patterns among the eleven most abundant species assessed in Florida.................................77

2.05

MDS showing the similarities in response patterns among the eleven most abundant species assessed in Florida in two-dimensional state space......................................................................................................78

2.06

MDS showing the similarities in response patterns among all 36 species assessed in Florida in two-dimensional state space...............................80

2.07

The log percent relative abundance of the species observed at the 200 transects assessed...................................................................................83

2.08

The distribution of the proportion of ranks over the 200 sites that the most dominant species, Montastrea faveolata, displayed.....................87

2.09

The proportion of ranks displayed by each functional group across the 200 transects surveyed...........................................................................89

2.10

The relationship between rank per functional group and total abundance per transect for the 200 transects surveyed............................................90

xxii

2.11

Relationship between percent coral cover of each functional group and environmental influence of island passes, (W)......................................93

2.12

Orthogonal relationship between average percent coral cover for each functional group and the total coral cover for the 20 reef sites surveyed on the Keyswide Coral Reef Expedition...............................................95

2.13

The same graphs illustrating the relationship between average percent coral cover for each functional group and the total coral cover for the 20 reef sites surveyed as in Figure 2.12, but with different scales on the yaxes........................................................................................................96

2.14

Relationship between species richness of all corals and environmental influence of island passes, W, at each reef site......................................98

2.15

Relationship between species richness and total coral cover across the 20 reef sites..........................................................................................100

2.16

Relationships between functional group richness under the four levels of membership constraint and the environmental gradient of W across reef sites...............................................................................................103

xxiii

2.17

Regression of functional group richness versus total coral cover for each site........................................................................................................106

2.18

Relationship between species richness of each functional group and environmental influence of island passes, W, at each reef site............109

2.19

Regressions of species richness for each functional group on total coral cover for each site................................................................................113

2.20

Percent cover for each of the 38 species recorded on the 20 reefs surveyed on the Keyswide Coral Reef Expedition..............................129

2.21

Functional group cover for each of the four predominant functional groups recorded on the 20 reefs surveyed on the Keyswide Coral Reef Expedition............................................................................................130

3.01

A photomosaic map of the Bermuda Islands and surrounding reef platform...............................................................................................134

3.02

An illustrated map of the islands and surrounding lagoonal patch reefs of Bermuda, with important geographic features labeled (produced by the author as part of the Bermuda Zoological Society).......................137

xxiv

3.03

A graph charting the number of passages by ships traveling through the southern shipping channel in 2004......................................................145

3.04

An aerial photograph of a cruise ship traversing the south shipping channel of Bermuda in a westerly direction, and leaving a plume of sediment in its wake............................................................................146

3.05

Temperature record for 2 subsurface temperature data loggers located either near North Shore (Inshore) or on the forereef at 30 ft depth (Offshore) for a three year period from 1998 – 2000 (modified from de Putron 2003)........................................................................................147

3.06

A graph of the hourly positions and paths the sun appears to take as it crosses the sky in Bermuda over the course of a day during the summer and winter solstices, and either equinox..............................................149

3.07

Zonal boundaries, locations of the north and south shipping channels, and location of patch reefs distributed across the study area encompassing the North Lagoon.........................................................153

3.08

A diagram illustrating the modified Li-Cor scalar PAR sensor...........159

xxv

3.09

Mean proportion of surface light (± standard error) originating from four directions at five depths, as measured by the hemispherical sensor .............................................................................................................160

3.10

Map of the five reef sites, indicated by the light-bulb symbol............163

3.11

Light intensity readings taken over five days from sensors positioned at 8-m depth at 5 reef sites located at different distances from shore across the area of study...................................................................................166

3.12

Average light intensity (Lumens ±SE) measured from 11 am to 1 pm local standard time over the first of the five days of deployment, by light sensors located at 8-m depth at five reef sites positioned at increasing distances from the North Shore of Bermuda......................167

3.13

Representation of how varying levels of the four most important environmental factors act to promote growth, or act as a stressor or disturbance agent to corals..................................................................171

4.01

A modified version of Grime’s (1977) and Steneck and Dethier’s (1994) generalized two-dimensional AST model of FG dominance within habitat types, incorporating the concession that biota can only survive in habitats within which the rate or amount of resource acquisition

xxvi

(resource abundance) is greater than the rate or amount of resource loss (or disturbance)....................................................................................177

4.02

General design of the study, in which survey sites (circles) were surveyed at a range of depths on the north, south and top sides of replicate patch reefs within each of six zones located at increasing distances from shore............................................................................183

4.03

A map of the lagoonal reefs located within and around the research area and the 18 patch reef sites surveyed in the videographic analyses......184

4.04

Diagram illustrating the average depths of each site on patch reefs located on different sides (aspects) and at varying distances from shore .............................................................................................................194

4.05

Dendrogram of Bray-Curtis similarities of species and functional groups clustered according by group-averaging..............................................196

4.06

Multidimensional scaling diagram (MDS) of square-root transformed relative abundance data for coral species averaged across sites located on replicate reefs and over different aspects, depths and distances from shore.....................................................................................................197

xxvii

4.07

MDS of square-root transformed relative abundance data of species for all sites.................................................................................................202

4.08

MDS of square-root transformed relative abundance data of species for all sites depths on all reefs...................................................................206

4.09

MDS of square-root transformed relative abundance data of species for all sites.................................................................................................209

4.10

The sites surveyed across the Bermuda platform cluster into three groups with different coral species composition, which also match three different environmental conditions......................................................214

4.11

MDS of square-root transformed frequency of occurrence data of all coral species for all sites, as in the three figures abov.........................216

4.12

The average proportion of frames with any coral present across all sites, illustrated as a line graph per site per reef (A) and as a bubble graph per depth and zone (B)...............................................................................217

4.13

Four MDS graphs of square-root transformed frequency of occurrence data of Branched Viviparous species as a group, and for M. decactis, M. mirabilis and P. porites corals separately for all sites.........................219

xxviii

4.14

The average proportion of frames with corals of the Viviparous Branching (VB) functional group present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom).......................................................................................220

4.15

The average proportion of frames with corals of the species Madracis decactis present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom)............221

4.16

The average proportion of frames with corals of the species Madracis mirabilis present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom)............222

4.17

The average proportion of frames with corals of the species Porites porites present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom)............223

4.18

MDS graph of square-root transformed frequency of occurrence data of the coral species Agaricia fragilis, the only member of the Foliose and Plating Viviparous (FP) functional group observed in the North lagoon in Bermuda in this study......................................................................225

xxix

4.19

The average occurrence frequency of the species Agaricia fragilis across all sites, illustrated as a line graph per site and reef (top) and as a bubble graph per depth and zone (bottom...........................................226

4.20

Four MDS graphs of square-root transformed data of occurrence frequency by Massive Viviparous species as a group, and for F. fragum, P. astreoides and S. radians corals separately, for all sites..................228

4.21

The average proportion of occupied frames per transect with corals belonging to the Massive Viviparous functional group present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom)......................................................229

4.22

The average proportion of frames with corals of the species Favia fragum present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom)............230

4.23

The average proportion of frames with corals of the species Favia fragum present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom)............231

4.24

The average proportion of frames with corals of the species Siderasterea radians present across all sites, illustrated as a line graph

xxx

per site per reef (top) and as a bubble graph per depth and zone (bottom)...............................................................................................232

4.25

Seven MDS graphs of square-root transformed relative abundance data of (a) Massive Oviparous species as a group, and for (b) M. cavernosa, (c) M. faveolata, (d) M. frankesi, (e) D. labyrinthiformis, (f) D. stigosa and (g) S. intersepta separately for all sites.........................................234

4.26

The average proportion of frames with corals of the Massive Oviparous (MO) functional group present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom)...............................................................................................235

4.27

The average proportion of frames with corals of the species Montastraea cavernosa present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom)...............................................................................................236

4.28

The average proportion of frames with corals of the species Montastraea faveolata present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom)...............................................................................................237

xxxi

4.29

The average proportion of frames with corals of the species Montastraea franksi present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom)...............................................................................................238

4.30

The average proportion of frames with corals of the species Diploria labyrithiformis present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom)......239

4.31

The average proportion of frames with corals of the species Diploria strigosa present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom)............240

4.32

The average proportion of frames with corals of the species Stephanocoenia intersepta present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom)...............................................................................................241

4.33

Average percent coral cover for (A), (B) species richness, (C) functional group richness as well as (D – F) the average percent cover for each functional group on the tops of each of three replicate reef sites in each of the six zones....................................................................................246

xxxii

4.34

Percent cover, species richness and functional group richness of corals surveyed on sites located on the tops and southern flanks of patch reef .............................................................................................................254

4.35

Percent cover of the Branched Viviparous, Massive Viviparous and Massive Oviparous functional groups of corals surveyed on sites located on the tops and southern flanks of patch reefs........................254

4.36

Percent cover of the three species of Branched Viviparous functional group....................................................................................................257

4.37

Percent cover of Agaricia fragilis, the one species of the Foliose and Plating functional group found within the lagoonal sites surveyed.....258

4.38

Percent cover of the three species of Massive Viviparous functional group....................................................................................................259

4.39

Percent cover of the two of the five species of Massive Oviparous functional group...................................................................................260

4.40

Percent cover of the three other species of Massive Oviparous functional group...................................................................................261

xxxiii

5.01

Diagrams depicting the differing ways in which the abundances of competitive (C), stress-tolerant (S) and ruderal (R) functional groups of corals are predicted to vary across habitats located across the range of stress and disturbance gradients encompassed by the AST model, and depending on the degree of niche overlap exhibited by each functional group....................................................................................................272

5.02

A diagram illustrating how the Zero Net Growth Intercepts (ZNGI) of each of the predominant functional groups of Caribbean coral found in Florida and Bermuda are dispersed across the Adaptive Strategies Theory model.......................................................................................273

5.03

The bounded area laid over the modified Adaptive Strategies Theory shown in Figure 5.02 represents the range of habitat types surveyed in Florida..................................................................................................274

5.04

The bounded area laid over the modified Adaptive Strategies Theory shown in Figure 5.02 represents the range of habitat types surveyed in Bermuda...............................................................................................275

5.05

A network of interacting corals on the Bermuda fore reef...................278

xxxiv

ABSTRACT Murdoch, Thaddeus James Thomas, Ph.D., University of South Alabama, December 2007. A Functional Group Approach for Predicting the Composition of Hard Coral Assemblages in Florida and Bermuda. Chair of Committee: Dr. Richard B. Aronson In Florida the functional-group approach provided new insights into the manner in which varying levels of disturbance affected species richness across sites. Despite the chaotic patterns in biomass displayed by each assemblage of coral species when separately plotted across reefs, each functional group of corals responded to direct and indirect gradients of disturbance in a orderly and group-specific manner. Functional groups displayed a nested distributional pattern, indicating that negative interactions between functional groups are probably weak Terrestrial and marine ecologists have found that a functional group approach can accurately predict how organisms will respond to changes in environment conditions. A functional group approach categorizes organisms, regardless of phylogeny, according to similarities and differences in life history and other ecologically relevant traits. One such model, the "CSR plant strategy theory" developed by Phillip Grime in 1973 for terrestrial plants, predicts the assemblage structure of biota over gradients of stress and disturbance. To test the CSR model, coral assemblages on reefs from Florida and Bermuda were assessed at the hierarchical levels of species and functional groups. The data were used to address the question of whether the functional-level approach provides information about community structure that species-level analysis fails to provide.

xxxv

Additionally, the predictions of the CSR model were tested regarding how coral cover, species diversity and assemblage structure should vary in habitats characterized by differing levels of disturbance and resource-limitation. In Bermuda, functional groups of corals also displayed a nested pattern across sites located over a range of depths and reef zones. When species were aggregated according to shared habitat, species from the same genus co-occurred in almost every case. This implies that these closely related species also share many functional traits and yet still coexist in many habitats The Adaptive Strategies Theory provides a series of simple, testable hypotheses that can be used to guide ecological research in an iterative and informative manner. The Adaptive Strategies Theory is a powerful theoretical framework, which can be modified to give it great heuristic value for guiding ecological research.

xxxvi

CHAPTER 1: THE NEED FOR A FUNCTIONAL GROUP APPROACH TO DESCRIBE THE STRUCTURE OF CARIBBEAN HARD CORAL ASSEMBLAGES Introduction Coral reefs are in decline around the globe (Aronson and Precht 2001, Knowlton 2001, Gardner et al 2003, Wilkinson 2004). Concerns that assemblages of reef corals may have lost their ability to resist disturbance are mounting (Jackson et al 2001; Nyström and Folke 2001; McClannahan et al 2002; Bellwood et al 2004; Hughes et al 2005; Aronson and Precht 2006; Nyström 2006), as exhibited by dramatic changes in the community structure of coral reefs, from a state of high coral biomass and low algal biomass to an alternate condition of high algal biomass and low coral biomass Gardner et al. 2003. In the Caribbean, the most obvious change is the loss of the three dominant species of coral (Acropora cervicornis, A. palmata and Montastraea annularis species complex), and the clear zonation patterns these species once produced (Done 1983; Graus and Macintyre 1989; Jackson 1991; Hughes 1994). Nonetheless, on many of the same reefs, it appears that subordinate coral species have not decreased to the same extent (e.g. Bak and Engel 1979; Aronson and Precht 1999). While it is well known that corals differ in their sensitivities to a range of environmental and biological factors, coral ecology as a science does not yet provide the means for predicting which species will be affected by specific changes in their environment nor in the manner in which changes will manifest themselves.

1

If we are to effectively manage coral reef communities and prevent further declines of corals, we must improve our understanding of how all species of coral respond to changes in physical and biological processes. We need to be able to determine whether predictable patterns in the abundance, co-occurrence and diversity of corals occur across habitats that vary in environment or in disturbance history. We also must ascertain the role that all species of corals play in providing function to reef ecosystems, including reef growth, nutrient cycling, the inhibition of invasive species, and the enhancement of biodiversity. With the exception of what historically were some of the more abundant coral species, such as Acropora cervicornis, Montastraea annularis and Porites astreoides, there has been very little advancement in these areas. Expanding our focus to all of the corals that live on a coral reef will require developing techniques for simplifying the complicated data that accompany such an increase in perceptional scope. The degree of complexity inherent in multi-species data from a large-scale ecological assessment can be seen in the following example. During the Keyswide Coral Reef Expedition of 1995 (Murdoch and Aronson 1999), 19,055 individual corals, representing 38 species, were recorded from 200 video transects filmed on twenty reefs located across the entire 350-km long Florida Reef Tract. When the percent cover data for each species on each reef surveyed was plotted (Figure 1.01), the resulting graph appears to have little structure. Instead the graph may best be described as a chaotic tangle of species varying in occurrence across reefs in an idiosyncratic manner. The lack of pattern found in data from Florida is typical for monitoring projects that cover large regions (e.g. Goreau 1959; Done 1982), and

2

illustrates the need for techniques with which to organize and simplify species-level data so that they can be interpreted in an ecologically meaningful way.

Figure 1.01. Percent cover for each of the 38 species recorded on the 20 reefs surveyed on the Keyswide Coral Reef Expedition. Letters represent individual sites on reefs separated by ~ 10 km. Data from each reef are plotted from westernmost to eastern-most location. Recently it has been suggested that a revived focus on the biological traits of organisms and the manner in which they vary across environmental gradients will promote the development of better ways of measuring and interpreting ecological information (McGill et al. 2006). Biological traits are specific, quantifiable characteristics of an organism that can be compared across individuals both within and among species.

3

Gradient analysis can include either indirect gradients, such as differences in depth down a fore reef, or direct gradients of a particular physical or chemical parameter, such as light intensity or oxygen concentration. A trait-based approach uses the properties of environmental gradients to tease apart the relative form that functionally important traits take in an assemblage of organisms located across the gradients, in much the same manner that glass prisms can be used to split a beam of white light into its component colors (Keddy 1992).

Functional Traits and Functional Groups in Reef Corals Corals are affected by and react to environmental and biological factors through their physiological, morphological and behavioral traits. Within a geographic region, the members of the species pool that are found within a particular habitat are presumed to possess the traits that allow their recruitment and continued presence, whereas the species that are absent are assumed to lack these same critical characteristics (Bradbury and Loya 1978; Sorokin 1993; Sullivan and Chiappone 1993; Edinger and Risk 1995; Hughes et al. 1999). Many researchers have attempted to detect theoretically meaningful correlations between the traits that different coral species possess and the species’ abundance within a habitat. For instance, Lang (1973) looked for patterns in competitive dominance hierarchies of corals by ranking species according to their aggressive abilities. Porter (1976) and Green et al. (1987) grouped species by polyp size in order to ascertain whether species with similar polyps shared trophic position and thus depth zones on forereefs. Szmant (1986), Edinger and Risk (1995), Hughes et al. (1999), Knowlton (2001) and others have grouped corals by reproductive mode. Corals that brood planula

4

larvae were compared with corals that broadcast spawn, and predictions regarding their relative abundance or distribution were made based on the differences in energy allocation, dispersal method and dispersal range between brooders and spawners. Barnes (1973), Jackson (1979), Bellwood et al. (2004), and others grouped corals and other reef biota according to morphology. They theorized that, since morphology is related to the rates of light collection, sediment shedding and fragmentation, morphology could be used to predict which coral species would dominate a particular habitat. However, all of these attempts at using single kinds of traits to predict the presence or abundance of corals in a particular habitat have been unsuccessful. This is because a single-trait approach cannot define differences between all species, not can it encompass the broad range of energetic, physical, chemical and biological factors and processes that affect corals on reefs. A more powerful technique is to examine how sets of different kinds of co-occurring traits are correlated with, or induced in response to, differences among habitats located along environmental and biological gradients (Keddy 1992; Körner 1993)). The manner in which species share groups of traits may be analyzed in two ways: (1) Functional trait analysis (2) Functional group analysis Functional trait analysis looks at the form each trait takes separately at the hierarchical level below the level of the species. Direct analysis of traits has been shown to be a powerful technique for interpreting the causes of the spatial distributions of terrestrial plants (e.g. Weiher et al. 1998; Mayfield et al. 2006), but it also further increases the complexity and amount of information needed. Alternatively, in functional group analysis, species that share life history or adaptive strategy are sorted into functional

5

groups, a hierarchical level above species. Functional group analysis has been shown to be useful for interpreting terrestrial plant data, but it provides the added benefit of simplifying species-level information, not making it more complex (Fagerstromm 1988; Körner 1993).

Assigning Species to Functional Groups using the Adaptive Strategies Theory Ecologists use functional classification schemes for two separate reasons (Gitay and Noble 1997). When investigating the effects of organisms on ecosystem processes, species are categorized into functional effects groups (e.g. Walker et al. 1999). Alternatively, when the goal is to determine the manner in which organism will react to environmental change, species are categorized into functional response groups (e.g. Lavorel et al. 1997). The research described in this manuscript focuses on the patterns manifested by functional response groups at locations that vary in environmental condition. There are as many ways to assign species to functional response groups as there are ecological factors of interest (Körner 1993). However, I propose that one particularly constructive way of classifying modular, sessile organisms such as plants, and perhaps corals, into functional groups is by using a modified version of Grime’s (1979) Adaptive Strategies Theory (AST; Keddy 1992; Andersen 1995; Steneck and Dethier 1996; Airoldi 1998). Grime (1979) used first principles to categorize all habitats into four primary kinds (Figure1.02), which I will refer to as habitat types, according to the relative measure within each habitat of two fundamental environmental factors. These two environmental conditions are (1) the availability or supply rate of resources (i.e. nutrients,

6

energy) employable for biomass maintenance and growth, and (2) the likelihood or rate that biota within a habitat will sustain damage or the loss of resources, biomass or physiological function (which is defined as disturbance). Following Grime’s (1979) scheme, the reciprocal of resource availability is defined as “stress”, and is considered to be different from disturbance. Accordingly, stress is defined as a lack of resources available for use by an organism, and disturbance is defined as a loss of the resources already acquired by the organism under consideration (Grime 1979), . Other terrestrial (Wilson and Keddy 1986; Chapin 1991) and marine (Kautsky and Kautsky 1989; Steneck and Dethier 1996) ecologists utilize the same convention, although the distinction between stress and disturbance is not usually made by coral ecologists (e.g. Dollar 1981; Grigg 1995; Hughes and Connell 1999) The specific suite of environmental conditions in which an organism finds itself affects the relative benefits and costs of allocating resources to different biological functions. Additionally, since the organism often finds itself in an environment with limiting resources, it must balance, or tradeoff, the amount of resources allocated to each function, based on the current adaptive value of that function, and relative to the adaptive value of the other functional structures and behaviors in which it could also invest. The organism must also minimize the risks inherent in allocating resources to a functional structure that has a high probability of being damaged or made redundant. All organisms must allocate resources to the following biological functions and behaviors: •

Resource acquisition

•

Maintenance and repair of body function

7

•

Resource and energy storage

•

Defense: The development of biochemical, structural or behavioral characteristics that prevent or inhibit other organisms from taking part of its body structure or resources.

•

Aggression: The development of biochemical, structural or behavioral characteristics that which facilitate competitive superiority for space, or other kinds of resource, over other organisms.

•

Growth

•

Sexual reproduction

•

Recycling of damaged or obsolete body structures and organelles.

Additionally, clonal organisms such as corals can allocate resources to asexual reproduction in any of three ways. First, they can reproduce asexually via fragmentation of viable parts of the colony, the rate of which dependent on the rate of growth and growth form of the species (Highsmith 1987). Second, some corals, such as Porites astreoides, are also capable of asexual reproduction by self-fertilization (Brazeau et al. 1998). Third, some corals, such as Pocillopora, may produce planulae asexually (Stoddart 1983; Sherman et al. 2006). Of the ten ways to allocate resources described above, three in particular play a key role in determining the life-history and functional characteristics of an organism and its survival abilities in different environmental conditions (Table 1; Grime 1973). These three primary processes to which an organisms must allocate resources in order to persist within a habitat are:

8

(1) Growth, (2) Defense and Resource Storage, and (3) Sexual Reproduction. Within habitats characterized by high levels of available resource, and a low risk of disturbance (Figure 1.02), ecologically successful organisms will be those that primarily allocate resources to growth. Similar rates of growth could not be supported in habitats with low levels of resource availability, even with low levels of disturbance, and survival in such an environment would instead require resource allocation primarily to storage and defensive structures and behaviors. Alternatively, organisms are likely to lose stored or growth-directed biomass in habitats characterized by high levels of resource, but high rates or intensity of disturbance. In these heavily disturbed environments, resources should primarily be allocated to reproduction, so that offspring may escape to lessdisturbed habitats. Following the logic of Grime (1973), no strategy exists that permits the survival of an organism under the concurrent conditions of intense disturbance and negligible resources for repair or reproduction.

9

Table 1.01. The characteristic differences in biological attributes between competitive, stress-tolerant and ruderal plant species (modified from Grime 1979). Adaptive Strategy Attribute

Competitive

Stress-tolerant

Ruderal

Maximum Size

Large

Small

Small

Longevity

Long or short

Very long

Very short

Reproductive Maturity

Late

Late

Early

Reproductive Effort

Small

Small

Large

Reproductive Method

Both

Clonal

Sexual

Growth Rate

Rapid

Slow

Rapid

Stress response

Rapid

Slow

Reproduces

Palatability

Variable

Low

High

10

Figure 1.02. A modified diagram of Grime’s (1979) Adaptive Strategy Theory for classifying habitats according to levels of stress and disturbance. The diagram also illustrates the optimal strategy predicted to be exhibited by the biota found within each habitat type, based on the optimal use of resources and the likelihood of incurring damage.

11

Characteristics of each Adaptive Strategy

Competitive dominant species Organisms that primarily utilize a strategy in which growth is favored are termed “competitive dominants” by Grime (1979; Figure 1.02). Allocation of resources to growth is an optimal strategy when resources are abundant and levels of disturbance are low, because growth promotes further resource capture and allows even more growth. This positive feedback loop allows competitive species to reach large sizes, when in benign habitats, relative to species employing other strategies. High rates of growth also permit competitive species to expand (laterally and vertically) more quickly than less competitive species, and thereby dominate previously unoccupied space. Occupied space can also be actively acquired using growth, via either the over-topping or shading-out of slower growing organisms. In benign habitats the storage of resources is disadvantageous for competitively-superior species, since stored resources cost resources and energy to store and also tie-up resources that would be better used in the acquisition of more space and more resources. Asexual reproduction via fragmentation or similar mechanisms, which Grime (1977) refers to as vegetative reproduction, is expected to be enhanced in competitive species, since high growth rates, coupled with of partial mortality, will result in the generation of disconnected clones of relatively large size. Alternatively, the proportion of resources used for sexual reproduction in competitive species is expected to be relatively small, since the release of gametes represents a risky loss of resources that could also be used for additional growth and acquisition of space (Williams 1975; Bazzazz et al. 1987; Hall

12

and Hughes 1996; Heino and Kaitala 1999). As disturbances occur in all habitats, however, some level of reproduction is necessary for survival by species using all strategies. Reproductive onset in competitive species is predicted to be delayed until the organisms have grown to a large size. When initiated, reproduction is expected to occur after the season of maximal potential productivity or during the season of highest likelihood of disturbance. When exposed to either periods of stress or of disturbance, competitive species are expected to further reduce the allocation of resources to reproduction or storage, in favor of continued growth or tissue repair (Grime 1979). Ruderal Species Organisms that allocate most resources to reproduction are defined by Grime as “ruderal” or weedy species. Ruderal species are those that are capable of surviving in habitats characterized by high levels of disturbance, but only when abundant resources are also available. Ruderals are predicted to have life history strategies that differ substantially from those of competitive species, except that they share in the ability to rapidly capture resources. Since the likelihood or intensity of disturbance is high for ruderal organisms, they are likely to experience high rates of partial or total mortality and rarely reach large sizes. Species that allocate resources to the development of structures or physiological attributes that reduce the effects of disturbance would be more likely to persist in disturbed environments, but at a cost in the amount of resources available for the development of other tissues or for reproduction. Initial growth rates may be high in ruderal species, but since the relative cost of reproduction outweighs the risk that reproduction will reduce survivorship in highly disturbed environments, ruderal species

13

are predicted to initiate reproductive effort at a small size. Resource storage would be disadvantageous in ruderal species since lifespan is likely to be short. Fragmentation rates are predicted to be high in ruderal species. However, since fragment size is likely to be smaller than in competitive species, the fragments of ruderal species may have a lower likelihood of survival. Under periods of stress, ruderals are expected to increase resource allocation to reproduction, since the likelihood of continued survival within the habitat is reduced. Stress-tolerant species Species that allocate resources predominantly to storage and defense are termed “stress-tolerant species” by Grime (1979). Since the rate or probability of acquisition of resources is low, stress-tolerant organisms should possess structures that maximize resource capture and storage when resources are present. Allocation of resources for biochemical, structural or behavioral modifications that reduce the loss of biomass by predation or competition would also be maximally advantageous under stressed conditions. Low rates of resource capture will limit growth and reproductive output and delay the initiation of reproduction in these organisms to “mast” years when resources are particularly high. However, despite slow rates of growth the eventual attainment of a large size would be possible if the organisms were located within a habitat experiencing very low levels of disturbance. Additionally, the slow rates of growth and low density of biomass in stressed habitats slow the rates of competition, allowing a high number of species to coexist (Huston 1994).

14

The Graphic Model of the Adaptive Strategy Theory In this section I describe in detail the graphic model of Grime’s (1979) Adaptive Strategy Theory. I then explain how it has been refined by various investigators since it was first proposed. I also offer my own modifications to the graphic model, with which I hope to improve both the model’s clarity and the way it logically corresponds with reality. As shown in Figure 1.02, above, the initial graphic model of the adaptive strategy theory was a square subdivided into four boxes, each representing an extreme in environmental condition possible within a habitat (Grime 1979). However, since the high stress and disturbance environment was defined as uninhabitable by all biota, Grime rearranged the square habitat model as a triangle, or ternary diagram. In this new configuration, competitive ability, intensity of disturbance and intensity of stress are represented by three axes (Figure 1.03). The C, S and R axes of the ternary graph are the source for a second name for the Adaptive Strategies Theory, which is the “CSR model”. Secondary and tertiary strategies, which represent compromises in adaptive traits between the three primary strategies, such as “Competitive-Ruderal”, are hypothesized to exist in habitats characterized by intermediate levels of stress or disturbance (Grime 1979).

15

Figure 1.03. The Adaptive Strategies Theory graphic model, also known as the CSR model, depicted as a ternary diagram. Primary functional strategies are represented as follows: C = competitively dominant, S = stress-tolerant, R = Ruderal. Secondary and tertiary strategies represent a compromise between two or three of the primary strategies, and are represented in the model by combinations of these three letters.

Grime (1977) added the variable “Competition” and modified the graphic model into a ternary diagram so that the trade-offs between the adaptive strategies that confer advantage under each of the three strategies could be represented illustratively in a simple and intuitive manner. However, one problem with the ternary version of the model is that it represents three variables constrained within two dimensions. To be more accurately represented, the three variables should be considered to be independent of each other, and thus each should have their own axes on a three-dimensional model (Loehle 1988). This error in the graphic model means that the relative overall cost of response to each species is constrained to the same level of cost as all other species under study. Such a graphic constraint is consistent with the role of trade-offs between the three adaptive strategies

16

that Grime intended. However, it is more likely that some species will be better or worse than others at acquiring resources or at balancing the trade-off between the intensity of interspecific competition, resource availability and disturbance level, or that the tradeoffs are not linear. Therefore, Loehle (1988) contended, to more accurately depict nature, each species should be located on points within a relatively bigger or smaller triangles in the CSR ternary model. Of course Grime’s intent was to produce a relatively simple model of all possible adaptive strategies, which Loehle’s proposed three-dimensional model would not be. More importantly, however, it could be argued that adding competition to the square graphic model to make it a ternary model mixes the two independent variables, representing the environmental condition of the habitats, with the dependent competitive response that the organisms within the habitats are predicted to make (Steneck and Dethier 1994; Wilson and Lee 2000). In order to avoid the problems associated with combining the dependent variable of competition and the two independent variables in the graphic AST model, Steneck and Dethier (1994) restructured the CSR ternary diagram of Grime (1977) back into a two dimensional graphic (Figure 1.04). This reconfiguration restores resource availability and disturbance to their capacity as independent variables, and C, S and R as dependent response variables within the environmental state space.

17

Figure 1.04. Generalized model of community dominants by Steneck and Dethier (1994) that refines to Grime’s (1977) AST model (shaded in light gray). Primary strategies are in accordance with the AST with the addition of disturbanceadapted biota in the upper right corner of the model, which Grime later refuted (Grime 1995). Secondary strategies are indicated by letter designations.

However, I contend that Steneck and Dethier (1994) made a different error in their graphic model (Figure 1.04). In their restructured model, Steneck and Dethier (1994) included as inhabitable, areas under the graph in which resources are at very low levels but in which disturbances are moderately high. If one assumes: (1) that the disturbance gradient represents a range of rates of resource loss, (2) the gradient of stress on the y axis represents a range of rates of resource gain, (3) that similar positions on the two axes are intended to represent comparable levels of resource flux (albeit opposing directions of flux), and (4) that habitat types in which rates of resource loss are greater than resource gain cannot support organisms, then logically one should conclude that locations on the graph representing a loss-gain ratio greater than 1 should be empty of functional groups. To encompass these assumptions and conclusion, I suggest that the graphic model should

18

be redrawn as in Figure 1.05. In this further revision of the AST/CSR graphic model, viable biological strategies are only shown to exist in the areas of state-space in which the rate of resource gain is greater than the rate of resource loss. The line defining the boundary between viable and intolerable environmental conditions represents the zero net growth intercept (ZNGI) of the entire assemblage under investigation, in a manner similar to that used by Tilman (1982; 1989), Chase and Leibold (2003) and others. With the revised graphic model one can better represent how changes across habitats in the environmental factors of resource availability and disturbance affect the characteristics of coral assemblages. In Chapter 2 I delineate how the AST predicts each functional group of corals will exhibit different levels of abundance, species richness and functional ecology on reefs located across natural or anthropogenic gradients of stress and disturbance.

19

Figure 1.05. A modified version of Grime’s (1977) and Steneck and Dethier’s (1994) generalized two-dimensional model of FG dominance within habitat types. This modified version incorporates the concession that biota can only survive in habitats within which the rate or amount of resource acquisition is greater than the rate or amount of resource loss. The boundary between the white and grey areas is the ZNGI of the assemblage as a whole. Letter designations for adaptive strategies are as in Figure 2.

Assigning Caribbean Reef Corals to Functional Groups and Defining the Critical Tests Functional classification schemes such as the AST have been most typically applied to the study of terrestrial and marine plants (e.g. Grime 1979; Steneck and Dethier 1994; reviewed in Solbrig 1994). Some animals, such as ants, also possess many of the same characteristics that allow plants to be grouped according to functional response, such as

20

modularity, a dispersive reproductive phase and a sessile adult phase, and competition for space, and for this reason have also been investigated under various functional classification schemes (Andersen 1995). Reef corals equipped with zooxanthellae also share many of the characteristics that plant ecologists use to differentiate functional groups in plants (Furla et al. 2005), and for this reason may also be meaningfully classified into functional groups using the sorting strategies developed by plant ecologists. Like plants, reef corals possess a modular structure and can be composed of an indeterminate number of repeating multicellular units. Corals adopt many of the same basic morphological shapes as plants, and these shapes typically share terminology, such as foliose, palmate or bushy. It is not important that corals do not produce the exact same morphologies as plant, just that corals and plants are both capable of utilizing their modular character to produce a wide range of morphologies that differ in functional effect and response to the environment and to competition. Corals and plants have comparable life histories, with a dispersive reproductive phase followed by a sessile adult phase. Both kinds of organisms also generally rely on light-driven photosynthesis and the acquisition of water-dissolved nutrients for the resources and energy needed for growth and other life-sustaining processes. Some corals and some plants are “ecosystem engineers” (Jones et al. 1994), which produce topographic complexity that provides habitat for other organisms, thereby enhancing the biological diversity of the habitats they occupy. Additionally, both corals and plants maintain their spatial position through interference competition with neighboring sessile organisms.

21

I contend that Caribbean reef coral species and the potential functional groups can be assigned to positions within Grime’s Adaptive Strategies model, through reference to the same kinds of attributes and characteristics in corals as Grime utilized for plants. Reef corals appear to possess sets of traits that are indicative of adaptive strategies similar to those defined for plants by Grime (1977). For example, Soong (1993) observed that different species of massive (in this case, meaning hemispherical or mound-shaped, following Budd et al. 2006) reef corals that shared maximum size (large or small) also exhibited similar rates of growth and recruitment, reproductive mode, reproductive season, and the size at which each species reached sexual maturity (Table 2). Specifically, large massive corals appeared to exhibit a more K-selected strategy, with a spawning reproductive mode, relatively high rates of colony growth as adults, delayed puberty and low reproductive investment. Conversely, small massive species of corals exhibited a more ruderal strategy, with a brooding reproductive mode, slower growth as adults, and higher investment in reproduction and recruitment. The same ranges of and trade-offs in trait values can be seen in Grime’s competitive and ruderal adaptive strategies for plants. These similarities between Soong’s (1993) data for corals and Grime’s (1979) plant groups imply that: 1) corals are also constrained in the manner in which they allocate resources to critical ecological functions and 2) additional functional groups of corals, defined according to other morphological and reproductive life history strategies, may match Grime’s other adaptive strategies.

22

Table 1.02. Life history characteristics of massive scleractinian corals as defined in Table 3 in Soong (1993). Character Reproductive Mode

Large Species Broadcasting

Small Species Brooding

References* 2, 11, 12

Reproductive Season

Once, Annual Cycle

Many Times, Lunar Cycle

11, 12

Puberty Size

Large

Small

13

Recruitment

Low? (sic)

High? (sic)

3, 5, 6, 7, 8, 10, 11, 12

Growth Rate High Low 1, 4, 9 _____ 1: Vaughan (1915); 2: Stimpson (1978); 3: Bak and Engel (1979); 4: Highsmith (1979); 5: Rylaarsdam (1980); 6: Rogers et al. (1984); 7: Fitzhardinge (1985); 8:Hughes and Jackson (1985); 9: Hudson (1985); 10: Wallace (1985); 11: Szmant (1986); 12: Soong (1991); 13: Soong 1993.

Johnson et al. (1995), in their investigation into the extinction selectivity of corals with different ecological and life history traits, also found that sets of traits covaried in Caribbean corals. In both extinct and extant corals, they found that branching species were significantly more likely to have small corallites and small colonies than other morphologies; massive corals were significantly more likely to have large corallites, large colonies, and be oviparous; and that plating corals were more likely to have intermediate sizes of corallites and be viviparous. Also oviparous corals were more likely to be gonochoric, while viviparous corals were more likely to be hermaphroditic; a relationship which was independently described by Carlon (1999). Hughes and Tanner (2000) noted similar characteristic differences between large and small, massive coral species as did Soong (1993). They found that Montastrea annularis,

23

a large massive spawning coral, displayed slower growth, was longer lived, and exhibited sporadic (seasonal) recruitment; all life history strategies of K-selected organisms. Conversely, Agaricia agaricites, a small brooding coral, has a shorter lifespan and more consistent recruitment, under even marginal conditions; both of which were recognized as ruderal strategies (Knowlton 2001). Edinger and Risk (1999) proposed the use of a reef classification scheme for Indonesian corals derived from Grime’s AST model. Their classification scheme primarily used coral morphology as a means of categorizing Indonesian corals into one of three adaptive strategies. In their scheme, conservation status of corals, equivalent to functional groups, were as follows: •

Competitive: Non-Acropora and foliose corals

•

Ruderal: Acropora species

•

Stress-tolerant: Massive and submassive corals

Edinger and Risk (2000) demonstrated that the conservation status could be predicted for Indonesian reefs by classifying them according to the relative dominance of these conservation classes. However, Edinger and Risk (2000) emphasized that their grouping strategy and conservation classes were designed explicitly for Indonesian coral reefs and that in other regions the categories should be changed to match regionally appropriate coral species and conservation goals.

Attributes used for Classification

24

The scheme I propose, morphology and reproductive mode are the primary traits used to define functional groups of coral species (Appendix 1). As indicated in Table 3, reproductive mode and morphology are the combined traits that indicate a broad suite of other functional characteristics of reef corals. The functional groups proposed are:

• Competitive dominant: Branched oviparous corals, • Competitive-Ruderal: Branched, viviparous corals • Ruderal: Massive, viviparous corals • Competitive – Stress-Tolerant: Massive, oviparous corals, • Stress-Tolerant: Plating, foliose and solitary corals, (only viviparous in the Caribbean).

25

Table 1.03. The ranking of each proposed functional group in ten critical traits, and the adaptive strategy to which they most closely represent. Smaller numbers represent higher rank and greater levels of each attribute. The morphological categories are: B: Branched; M: Massive; P: Plating and Solitary. The reproductive categories are: O:Oviparous; V: Viviparous. The reproductive methods are; F: Fragmentation; X: sexual reproduction. The adaptive strategies are: C: Competitive; CR: Competitive – Ruderal; CS: Competitive – Stress-tolerant: R: Ruderal; S: Stress-tolerant. Trait Maximum Size (Genet) Longevity (Ramet) Longevity (Genet) Reproductive Maturity Reproductive Effort Reproductive Method Growth Rate Stress Response Aggression Palatability Adaptive Strategy

BO 1 3 1 5 4 F>X 1 3 3 3 C

BV 3 4 2 2 2 F:X 2 4 4 2 CR

MO 2 2 3 4 3 F>X 3 2 5 4 CS

26

MV 5 5 5 1 1 F<X 4 5 2 1 R

PV 4 1 4 3 5 F<X 5 1 1 5 S

Review Reference Johnson et al. 1995 Hughes 1984 Highsmith 1987 Richmond 1998 Richmond 1998 Highsmith 1987 Huston 1985b Bak and Meesters 1998 Lang 1973 Rotjan and Lewis 2005

Characteristics of Each Functional Group of Coral Competitive dominant: Branched oviparous corals In the Caribbean, branched, oviparous corals are represented by Acropora and Oculina species (Appendix 1; Fadlallah 1983; Harrison et al. 1984; Richmond and Hunter 1990; Brooke and Young 2003). Corals of these two genera have high growth rates (Huston 1985b), indeterminate growth (Highsmith 1987) and typically dominate the surface of most coral reefs (Richmond and Hunter 1990; Brooke and Young 2003). Sexual reproduction occurs in late summer, the period when both resource availability and the risk of damage due to hurricanes are most likely, and thus when the risk of wasting resources via reproduction is offset the most (Woodley et al. 1981). Recruitment rates are typically very low (Bak and Engel 1979, Rogers et al. 1984, Harrison and Wallace 1990; Smith 1992, Richmond 1997), which confirms their status as competitive dominants which allocate resources to growth versus reproduction. These branched corals fragment easily and can apparently utilize this form of asexual reproduction to successfully disperse over small distances (Highsmith 1987, Lirman 2000). Their branched structure allows survival under high sediment loads, although episodic occurrences of high turbidity may inhibit rapid growth fueled by photosynthesis. Rapid growth and a tall, branched structure allows these corals to overgrow all other corals under benign environmental conditions. They are moderately aggressive in direct interactions with other corals (Lang 1973). While their skeletal structure provides moderate protection from predation by parrotfish, Acroporids are prone to corallivory by polychaetes (Woodley et al. 1981) and snails (Baums et al. 2003). The corals of this functional group should demonstrate low spatial variability among sites at one depth

27

within a reef, but high variability across areas of varying water quality (e.g. Lidz and Shinn 1991 ). Also, sections of the reef tract that receive more frequent or very intense disturbances, such as areas exposed to the open ocean, should have lower cover of branched spawning corals compared to less frequently disturbed areas (e.g. Geister 1977; De Meyer 1998; Parker and Oxenforn 1998).

28

Competitive-Ruderal: Branched, viviparous corals Branched corals of the genera Porites and Madracis, as well as the species Agaricia tenuifolia utilize a viviparous reproductive strategy (Appendix 1; Morse et al. 1988; Johnson et al. 1995; Richmond 1998). These corals typically have a digitate, branched, or morphologically plastic form (Smith 1984; Bruno and Edmunds 1996; Veron 2000) and can dominate large areas (Lewis and Snelgrove 1990; Chornesky 1991; Aronson et al. 1998). While the colonies can be very large, living tissue often does not connect neighboring branches (Smith1984; Lewis and Snelgrove 1990; Veron 2000), implying that branch longevity is much shorter than colony longevity. Branched viviparous corals release planulae over an extended period of time (Richmond 1997), resulting in relatively high rates of recruitment (Bak and Engel 1979; Smith 1992). Since they can also disperse through fragmentation (Highsmith 1987; Bruno and Edmunds 1997), corals of this functional group are able to take advantage of both asexual and sexual reproduction as a means of escaping disturbances and spreading across a reef (Bruno and Edmunds 1997). Branched viviparous corals exhibit weak aggression towards other coral species (Lang 1973). Additionally they are not structurally defended from corallivory, and can suffer high levels of predation by parrot fish (Murdoch, Looney and Aronson unpublished document; Grottoli-Everett and Wellington 1997, Miller and Hay 1998). The enhanced tolerance to disturbance and mix of reproductive strategies should allow branched viviparous corals to show moderate to high cover and low variability across reefs in marginal habitats (Aronson et al. 2005) , and low cover in areas where parrotfish or competitively dominant species are abundant.

29

Ruderal: Massive, viviparous corals Massive (mound-shaped), viviparous corals are found within a broad range of genera, including Agaricia, Favia, Manicinia and Porites (Appendix 1). Corals of these genera have smaller colony sizes (than massive, oviparous corals), determinate growth, a short lifespan, and high reproductive output (Soong and Lang 1992; Soong 1993). These corals tolerate greater extremes in disturbance than mound-shaped spawners (Sammarco 1985; Connell 1997), but are not good competitors for space (Lang 1973), and reproduce at a young age (Richmond 1990). Additionally many massive oviparous corals are capable of self-fertilization (Brazeau et al. 1998). As they reproduce continually over the year, less energy is available for growth (Hall and Hughes 1998). These factors indicate a ruderal lifestyle that is maximally adapted to frequent settlement and rapid growth within patches of marginal quality generated by disturbance. These corals are likely to be the first to settle in newly disturbed areas, and indeed may be the only corals present if conditions are extreme. Like viviparous branched corals, massive viviparous corals also are not well defended and are subject to high levels of corallivory by parrotfish (Rotjan and Lewis 2005). The corals of this functional group should demonstrate low variability among sites within reefs, as they have a wide range of environmental tolerance and high recruitment rates. Mound-like brooding corals should also show low variability from reef to reef, for the same reasons. As they are neither good competitors nor aggressive, ruderal corals should be less abundant than competitively dominant species on most reefs. Competitive – Stress tolerant: Massive oviparous corals

30

Massive oviparous corals include species from the genera Montastrea, Diploria, Colpophylia, as well as additional genera (Appendix 1). Corals that are mound-like and that spawn gametes generally possess moderate to high growth rates as adult colonies, indeterminate growth and can reach very large sizes (Soong 1993). As they only reproduce during part of the year, and generate gametes which need to be fertilized in the water column, they have moderate to low recruitment success (Smith 1992). Partial mortality may create large fragments with high survivorship. Massive oviparous corals are generally sensitive to sedimentation (Nugues and Roberts 2003), although this weakness may be offset by the corals’ ability to grow rapidly (Logan et al. 1994). However, their shape and size may aid in the survival of intense disturbances such as storms and hurricanes (Woodley et al. 1981, Liddell and Olhorst 1987). Massive oviparous corals have moderately large and plocoid corallites, which protect polyp tissue from predation by fish such as parrotfish. The corals of the massive oviparous group probably demonstrate high spatial variability in coral cover from reef to reef, depending on the water quality of each reef. Reefs in clear oligotrophic ocean water should have a high cover of large massive oviparous corals, with a large proportion of the population composed of competitively superior genets. Reefs in turbid or nutrient-rich water should have a lower cover of these corals, and the ones present should be smaller and more fragmented. The within-reef variability of this group of corals could be high or low, depending on the levels and history of disturbance, and age of the reef in question. Stress Tolerant: Plating, foliose, and solitary corals

31

Plating, foliose, and solitary corals, such as species within the genera Agaricia, Mycetophyllia and Scolymia, have slow growth rates and generally a moderate to small colony size (Johnson et al. 1995, Veron 2000). They tend to have thick tissue relative to skeletal thickness (Budd et al. 2006), and thus fragment rarely. However, since they produce brooded planulae (Johnson et al. 1995; Richmond 1998) which may be capable of settlement soon after release from the parent, recruits may tend to cluster in environments for which they have an adaptive advantage. The corals in this group generally rank high in Lang’s (1973) aggression scale, and often have complex skeletal structures, such as coarse septal dentition, (Budd et al. 2006) which may prevent complete tissue loss to polyps by either competition or predation. Stress-tolerant corals should be very patchily distributed on all scales, since they recruit near to the parent colony and are susceptible to disturbance and interference competition by most other types of coral.

Sources of Environmental Stress and Disturbance on Coral Reefs The four environmental factors that most strongly play a role in determining the characteristics of a coral assemblage at a reef site are temperature (Weber and White 1974; Glynn and Stewart 1973; Walker et al. 1982)), light (including UV light) (reviewed in Falkowski et al. 1990), current speed (Geister 1977), and suspended sediment load (Dodge and Vaisnys 1977; Acevedo et al. 1989; Anthony 1999). All four of these factors may be considered a source of enhanced growth or as a source of disturbance to corals, depending on the intensity or rate at which the coral is exposed to each factor.

32

Corals are adapted to a specific range of temperatures, and are negatively affected if exposed to temperatures that are only 2° above or below this range (Figure 1.06A; Walker et al. 1982; Glynn 1988). Cold temperature can be viewed as a stress, in that physiological processes are restricted by low temperatures. Conversely, high temperatures act as a disturbance by damaging the tissue of the coral host and also causing the expulsion of the symbiotic zooxanthellae in the process of coral bleaching (reviewed by Browne 1997). Corals are also adapted to a specific range of light conditions (Figure 1.06B), with excessive light causing damage to the both the coral host and also damage or the expulsion of their symbiotic zooxanthallae (Falkowski et al. 1990; Browne 1997). Since light is required by zooxanthallae for photosynthesis of carbohydrates, which are made available to the coral host, low levels of light result in a loss of resources to the coral colony. Suspended sediments (Figure 1.06C) negatively affect coral increasing the turbidity of the water column and there by blocking light transmission (McCarthy et al. 1974), smothering the coral polyps (Hubbard and Pocock 1972; Rogers 1990), or by abrading coral tissue (Rogers 1990). Alternatively, suspended sediments may provide nutrition to corals that are not otherwise available in oligotrophic waters (Anthony 1999; Mills et al. 2004) and corals may be nutrient stressed when both dissolved and particulate sources of nitrogen are in low concentration. Waves and other forms of water motion generate currents which affect corals in different ways depending on their strength (Figure 1.06D). Water motion is required to carry dissolved and particulate nutrients to the coral (Anthony 1999; Mills et al. 2004)

33

and low current speeds result in less nutrient resources being available to the coral colony due to the development of a boundary current. On the other hand, currents with a high rate of flow can break corals (Geister 1977; Tunnicliffe 1981) or suspend sediment and thereby abrade coral tissue (Rogers 1990).

34

Figure 1.06. Representation of how varying levels of the four most important environmental factors act to promote growth, or act as a stressor or disturbance agent to corals.

35

Testing the Applicability of the Functional Group Approach for Reef Corals In this dissertation, I test several hypotheses regarding how each different functional group of reef corals varies across environmental gradients of stress and disturbance. The specific predictions are as follows: Similarities in Distribution of Species Between and Among FG Species within functional groups are hypothesized to be redundant in terms of their functional responses to the environment. If this is so then species belonging to the same functional group should differ little in their distributions across sites, due to the shared responses to environmental and biological conditions (Steneck and Dethier 1994; Gitay and Noble 1997; Hooper et al. 2002). Conversely, the distribution patterns of species belonging to the functional groups should be substantially different due to disparate environmental tolerances. Under the above paradigm, species within environmental functional groups are hypothesized to be redundant in terms of their functional responses to the environment. If this is so then species belonging to the same functional group should differ little in their distributions across sites, due to the shared responses to environmental and biological conditions (i.e. Figure 1.07C). If, however, the traits used to distinguish species into functional groups are those primarily utilized for resource acquisition, and not environmental tolerance, then species within functional groups should be too similar to be able to persist within the same habitats (following Gause 1934). Under this scenario species that belong to the same resource-based functional groups (also termed “guilds”) should tend to occur in different habitats (reviewed in Fox 1999). Also species from different guilds should be able to

36

coexist to a greater degree than species from the same guild. Patterns of species distributions when grouped into guilds should match the pattern in Figure 1.07D, below. Conversely, if functional groups do not exist, or if the species within functional groups do not respond in unison to changes in environmental condition, then a serial (Figure 1.07A) or nested pattern (Figure 1.07B) of species turnover may occur across sites. Analysis of species abundance data, species presence-absence across sites, or the ordination of multidimensional species data for sites can be used to determine which of the above patterns are exhibits by corals in sites located across environmental gradients.

37

Figure 1.07. An illustration of different hypothetical distributions of species and functional groups across an environmental gradient (A to D), and how they appear when graphed as (i) abundances, (ii) tabulated in a matrix of presence vs. absence, and (iii) graphed using a multivariate ordination technique, such as Multidimensional Scaling (MDS). In the second column of figures, columns in the matrix represent sites and rows represent species and functional groups. Filled cells in a row represent the presence of a species and empty cells represent its absence. Each row represents a different hypothetical distribution, as follows: (A) The pattern of species presence when species and functional groups are replaced in series across an environmental gradient. (B) An alternate pattern, in which the range of sites inhabited by organisms with lower tolerances are nested within the range of sites occupied by more tolerant species and functional groups. (C) The distribution pattern in which functional groups exhibit turnover in response to the environmental gradient but species within functional groups coexist and have the same distributions (i.e. FG underdispersion). (D) A different pattern in which species only compete strongly within functional groups but not between functional groups. Under this scenario species of different functional groups are predicted to cooccur while species within functional groups distribute themselves with turnover across the environmental gradient (i.e. FG overdispersion)

38

Rank Abundance by Species and Functional Groups According to the AST (Grime 1979), species and functional groups should differ in dominance depending upon the levels of stress and disturbance of the habitat they are in. Alternately, the unified neutral theory (UNT) suggests (Hubbell 2001) that all coral

39

species are functionally equivalent, and the ranking of any species will be random from site to site, regardless of the environmental conditions of the reef in which it is found. If the paradigm of UNT is in operation, no species is expected to exhibit greater dominance when examined across a large number of sites. Conversely, it may be that species differ in dominance, but that no within-functional group similarities in dominance patterns exist. This would indicate that each species is different from all other species but that they do not share functional group affiliation as I previously designated them above. If, however, species from the same functional group tend to have similar dominance patterns across transects, this would indicate that functional groups exist and that species within functional groups tend to be fairly equivocal. In other words the UNT may apply within FG but not between FG. In Under this model, the pattern of dominance of species would conform with both the AST and a version of the UNT (Hubbell 2005). If the predictions of the AST are true, then the corals of the MO (massive oviparous) functional group should dominate on reefs with high colony density. This is because they are predicted to use growth to dominate when resources are abundant and levels of disturbance low. However, since low total colony abundance was shown above to be highly correlated with W, which I showed above is a proxy for disturbance, and the MO functional group is predicted to have low growth rates and survivorship when levels of disturbance or stress are high, it is predicted to not dominate on reefs with low total colony abundance. Conversely, species of the stress-tolerant and the ruderal functional groups, which are the foliose and plating functional group and the massive viviparous functional groups, are predicted to dominate reefs with low total colony density. The rank abundance of the species from each group will drive the rank abundance of each

40

functional group as a whole, and if this is the case then the MO functional group is predicted to dominate most reefs with the other functional groups occupying a subordinate position. Percent Cover of All Corals The first line in Grime’s (1977) classic paper defines stress and disturbance as “external factors limiting (plant) biomass”. In this scenario, the level of stress represents the relative rate of resource availability to organisms within a habitat, and disturbance the relative rate of resource removal or loss to organisms within a habitat. Habitats characterized by the presence of more resources (i.e. a high rate of resource gain), and fewer disturbances (i.e. a lower rate of resource loss), should exhibit more organisms, higher amounts of biomass or a higher percentage of cover by corals than habitats exposed to higher levels of disturbance or stress (Figure 1.08).

41

Figure 1.08. An illustration of how total assemblage biomass is predicted to vary across habitat types characterized by different levels of resource availability and disturbance. Percent Cover per Functional Group According to the AST, each functional group is adapted to survive only within a specific range of disturbance, stress and competition, representing that functional groups’ environmental and biological niche. If functional groups are affected by the environment in this manner, each primary functional group (C, ST or R) will dominate one of the three environmental extremes (see Figure 1.09A below), and also will decline in abundance rapidly in habitats characterized by other environmental conditions. Also each intermediate strategy (such as C-R, S-R or C-S) should peak in abundance at a midpoint

42

between environmental extremes. The abundance of each functional group is also expected to display little overlap from HT to HT according to this model. More recent theoretical and empirical evidence (O’Neill et al. 1988; O’Neill 1989; Peterson et al 1998; Bolker and Pacala 1999) indicates that the pattern in which each functional group heavily dominates one area within the AST model may not apply. Instead all functional groups may be capable of coexistence in almost all habitat types except the three extreme environments. If the functional groups that are not competitively dominant utilize either (1) recently cleared patches of high resource availability via source-sink dynamics (ruderals), or (2) patches of low resource quality (i.e. heavily shaded; stress-tolerant species), then they could persist in habitats exposed to low disturbance and high stress, despite the spatial dominance of the competitive group (Bolker and Pacala 1999). If these adaptive strategies are in effect, then ruderal and stress-tolerant species may also reach maximum cover in habitats with highest resource availability, and the overlap between functional groups will be high (Figure 1.09B and C below).

43

Figure 1.09. Diagrams depicting the differing ways in which the abundances of competitive (C), stress-tolerant (S) and ruderal (R) functional groups of corals are predicted to vary across habitats located across the range of stress and disturbance gradients encompassed by the AST model. Each model illustrates a different degree of niche overlap that may plausibly be exhibited by each functional group. In all graphs X and Y represent graphs of abundance relative to levels of disturbance (X) or stress (Y). Z represents the modified CSR square diagram, with the zero net growth intercept (ZNGI) illustrated for each functional group. Graph A matches Grime’s original predictions in which functional groups are limited to specific regions of adaptive niche phase space with no overlap in niche boundaries. Graphs B and C represent alternate models in which competitive species do not negatively interact with stresstolerant or ruderal species. In graph B the functional groups exhibit minimal overlap in niche boundary, whereas in graph C the functional groups exhibit maximal overlap in niche boundary. In all models C, S an R functional groups maintain dominance under differing environmental conditions.

Regardless of whether the niche model or the competitive hierarchy model is the more accurate description of reality, and irrespective of the level of overlap between species, the following predictions can be made: •

The competitive group will exhibit it’s highest level of cover at the least disturbed and least stressed site.

•

The competitive group will exhibit higher cover than all other groups at the least disturbed and least stressed site.

44

•

Other functional groups may peak in cover at some moderate level of disturbance or stress.

•

All functional groups should display lowest cover at the most disturbed or stress site.

•

The C, CS, and S groups should decline in cover more rapidly than R groups across gradients of disturbance. The S group should decline to minimal levels before the C group across a disturbance gradient.

•

C, CR and R groups should decline in cover more rapidly than S groups across gradients of stress. Total Species Richness

As previously stated in the Intermediate Disturbance Hypothesis (Grime 1973; Connell 1978; Sousa 1979; Aronson and Precht 1995; Dial and Roughgarden 1998 and many more), the number of species found within certain kinds of habitat will be limited by both environmental conditions and biological interactions. High levels of stress or disturbance should restrict species richness by limiting the rates of growth or by depleting populations faster than they can recover (Figure 1.10). Under low levels of stress or disturbance, species with the best ability to acquire resources and procure territory from other species will inevitably maintain a competitive advantage. The monopolization of space by these competitive species will obstruct the settlement and growth of subordinate species within patches, thus limiting the number of species within habitats that have low levels of stress and disturbance. For these reasons, in Grime’s Adaptive Strategies model (1977) a peak in species richness is expected to appear at the location within the model

45

where rates of population growth and rates of recovery from disturbance are both at moderate levels (Figure 1.10).

Figure 1.10. The distribution of species richness predicted to occur across habitats by Grime’s (1977) Adaptive Strategies model. High represents a peak in species richness at the midpoint between resource loss and gain. Low represents areas of reduced species richness due to environmental or biological factors. Circles enclose areas of equal species diversity.

Functional group richness Functional group richness represents the number of functional groups found within a particular habitat. While Grime (1977) does not make any explicit predictions about how functional group richness or diversity varies across his ternary model, he does plot the general location of different plant groups, such as trees, shrubs and herbs, within the

46

ternary diagram (Figure 1.11). In his illustration, the greatest number of plant functional types occurs at the midpoints of the two axes of resource flux, with fewer plant types found towards the edges. Unfortunately, Grime (1977) does not provide a theoretical reason for the pattern he predicts to occur. Alternatively, Steneck and Dethier (1994) predict that the number of functional groups will peak across habitat types in the same manner as does total biomass of the assemblage across the adaptive strategies model (Figure 1.12). Their pattern is based on the empirical results of their research into the distribution of marine plant functional groups across tropical and temperate sub-tidal habitats. In their graphical model the most stress- and disturbance-tolerant species are grouped into one inclusive category, and are located across all survivable habitat types. Additional functional groups then appear towards the corner of the square in which resources are not limited and disturbance levels are low. This pattern directly contradicts the general predictions of the Adaptive Strategies Model (Grime 1979), where ruderal species are defined as intolerant to stress, and stress-tolerant species are defined as incapable of persisting under heightened disturbance. Another problem with the Steneck and Dethier (1994) model is that no theoretical explanation for their predicted (nested) pattern is provided either.

47

Figure 1.11. A diagram illustrating the range of strategies encompasses by (a) annual herbs, (b) biennial herbs, (c) perennial herbs and ferns, (d) trees and shrubs, (e) lichens and (f) bryophytes. For the distribution of strategies within the model see Figure 4. Modified from Grime (1977).

48

Figure1.12. A three dimensional model of the levels of biomass predicted for set of functional groups of species across habitat types characterized by varying levels of disturbance potential and productivity potential. Modified from Steneck and Dethier (1996).

As stated above, there are considerable shortcomings with both of the above models. The prediction of Steneck and Dethier (1994) of a peak in the Competition section of the ternary model is counter that of Grime (1977), who states that the peak in functional diversity should occur in habitats experiencing the lowest levels of competition. Neither theory incorporates a prediction of how the richness or diversity of species within each functional group will vary across habitat types. Finally, both models lack a theoretical underpinning. In order to address these issues, in the next paragraph I put forward a theoretical explanation for how the components of (1) functional group diversity, and (2) the diversity of species within each functional group will vary across habitats of varying resource flux, as illustrated in the Adaptive Strategies Model.

49

In the restructured model I propose, I define positions within the graphical Adaptive Strategies Model as habitats, in which each habitat is composed of a patchwork of habitat types that differ slightly in environmental condition (Figure 1.13; Connell and Keough 1985). Each position within the model defines the upper limit in the rate or amount of resource flux (gain or loss) within each habitat. In this manner, highly stressed habitats should only contain a small range of resource availability due to a low rate of resource gain, disturbed habitats should contain a small range of resource availability due to high rates of resource loss outpacing resource gain, while benign habitats should possess patches exhibiting a wide range of resource availability, due to resource gain outpacing resource loss.. Following this reasoning, stressed or disturbed habitats should be limited in the range of species and functional groups they can harbor, while benign habitat types should exhibit the full complement of functional groups. More functional groups are predicted to survive within benign habitat types despite the dominance of space by competitive dominants because some patches are likely to exist that possess lower levels of resource availability or higher levels of disturbance which the competitors cannot tolerate. Accordingly, functional diversity should peak in the upper left corner of the model, as it does in the model of Steneck and Dethier (1994), and not as predicted by Grime (1977). However, in the proposed model, unlike that of Steneck and Dethier (1994), ruderal functional groups are not predicted to occur within stressed habitats, and stress-tolerant functional groups are predicted to be absent from habitats subject to high levels of disturbance. Such a pattern is in accordance with the original Adaptive Strategy Theory (Grime 1977).

50

Figure 1.13. A diagram illustrating how functional groups are predicted to be dispersed across patches located with habitats defined by varying rates of resource gain and loss. Larger squares represent habitats. The smaller nested squares represent patches within habitats. Letters represent the functional group occupying each patch. C represents the competitive functional group. S represents the stress-tolerant group and R represents ruderals. Grey squares represent empty patches. The black field on the lower right side of the diagram represents the range of habitats in which high relative rates of resource loss limit biomass.

51

Species richness within functional groups The differences in life history strategy of each functional group should affect how the species within each group interact across gradients of stress or disturbance. Competitively dominant organisms are predicted to compete heavily under benign environmental conditions, but it may be that competition is only directed towards other members of the competitive functional group and not members of the other functional groups. Ruderal species are typically the poorest competitors for space, in both direct contact and overgrowth competition. Stress-tolerant species are characterized as the strongest defenders of space, but also recruit to patches that possess low levels of resource that could potentially reduce their rates of competition substantially. If it is true that competitive dominant species only compete with each other, then the species richness of the competitive group should be reduced in habitats where resources are more abundant and where disturbance levels are low. Under this scenario, the stress-tolerant and ruderal species would be expected to display highest levels of species richness in low stress and disturbance patches. Alternately if the species within the competitive functional group does compete against all other species regardless of life-history, or if species within each functional group compete strongly with other members of the same group, than all FG will display a reduction in species as resources increase and disturbance levels decline. Regardless of the potential differences in competitive interaction between functional groups, all functional groups are predicted to display a reduction in species richness in habitats characterized by low resource levels or high disturbance. The species richness of the competitive dominant and stress-tolerant functional groups should decline faster than the ruderal group’s species richness as disturbance levels increase, and the species

52

richness of the competitive dominant and ruderal functional groups should decline faster than stress-tolerant functional group’s species richness as resources levels decline across sites.

Testing the adaptive strategies theory on Caribbean coral reefs In the chapters to follow, I will examine the ability of the proposed functional group approach to predict the structure of coral assemblages at sites on reefs located across the environmental and biological gradients found across the Florida Keys reef tract and the Bermuda reef platform. In Chapter 2, I test the predictions of the adaptive strategies model with coral data from reefs in Florida. The Florida data are from deep fore reef sites located at 13 to 19 m depth. These reefs are characterized by a spur-and-groove geomorphology and separated by deeper, sandy areas. An earlier investigation determined that the Florida Reef tract displays distinctive scales of variability in total coral cover and abundance. Chapters 3 and 4 describe how I collected data from sites located at multiple aspects (i.e. compass bearing) and depths from patch reefs located at intervals across the lagoon that is found north of the island of Bermuda. By selecting sites that vary in depth and distance from shore, the responses of species and functional groups of coral to two distinct environmental gradients can be examined. In Chapter 5 I summarize the conclusions of the previous chapters and discuss their implications.

53

CHAPTER 2: THE RESPONSES OF FUNCTIONAL GROUPS OF CORALS TO DIRECT AND INDIRECT GRADIENTS ON THE FLORIDA REEF TRACT

Introduction In this chapter I examine how functional groups of corals are distributed across a gradient of disturbance (i.e., along the C-R axis of the AST model; Figure 2.01), over which the range of the resource (stress) gradient is at a minimum. This was done by comparing coral assemblages located at a single depth across reef sites known to vary in water quality. Standardizing the environmental conditions in this way allowed the examination of the hypothesized competition-colonization trade-off displayed between competitive and ruderal corals, while limiting the expected range of responses by the stress-tolerant functional groups. The 1995 Keyswide Coral Reef Expedition was a multidisciplinary survey of coral reef habitats located along the entire Florida Reef Tract. The results and mapping these coral assemblages showed that coral cover, colony abundance and diversity varied among reefs on deeper forereef habitats (13-19 m depth) (Aronson and Murdoch 1996; Murdoch 1998; Murdoch and Aronson 1999). Each biological measure (i.e. cover, abundance, diversity) showed little variability across sites separated by 1 km within reefs, demonstrated high variability among adjacent reefs separated by less than 10 km, and showed little variability

54

across regions sampled at the 50-km scale. Furthermore, focused analysis in the Middle and Lower Keys regions determined that exposure to inimical water from Florida Bay reaching the reefs through passes between the keys was a likely source of the high variability in community structure observed from reef to reef (Murdoch 1998). Florida Bay water does not flow into the other regions of the Florida Keys where reefs were assessed by Murdoch (1998), and so only reefs within the Middle and Lower Keys were including in the focused analysis. Florida Bay water possesses several characteristics that can limit coral growth or survival. These characteristics include extreme variability in temperature and salinity, and high nutrient and sediment loads (Shinn et al. 1989, Szmant and Forrester 1996). To test the hypothesis that Florida Bay water negatively affected the coral assemblages on spur-and-groove reefs in the Middle and Lower Keys, the average coral cover, as surveyed in the Keyswide Coral Reef Expedition (Murdoch and Aronson 1999), was compared with a dependent variable (W) by Murdoch (1998). The variable W was derived from (A) the average flow rate of Florida Bay water traveling through individual passes, which was measured by Smith (1994), divided by (B) the linear distance to the nearest up-current pass in km, as determined by maritime charts: W = flow rate (m sec-1) / linear distance from pass to reef (km)

55

(1)

Figure 2.01. The elongated oval within this square diagram of state space represents the hypothetical range within the AST (CSR) model that was occupied by the sites of the Florida Reef Tract that are the focus of this chapter.

56

Visual inspection of the graph and regression analysis of average coral cover at each site with W (Figure 2.02) indicated that coral cover declined in a linear fashion with an increase in W (i.e., an increase in flow rates or a decrease in the distance to the nearest pass). The least squares regression analysis of data describing the change in coral cover with change in W produced the equation: Coral cover = 17.381 - 0.2367(W)

(2)

The independent variable W (flow rate/distance to pass) explained much of the variance in coral cover. A t-test determined that the slope of the best-fit line differed highly significantly from zero [r2 = 0.8318, df = 9; t-ratio = -6.29, p = 0.0002]. The existence of a significant linear relationship between the variable W and coral cover on the reefs of the Middle and Lower Keys provides support for the conclusion that there exists a steep gradient of environmental factors in operation across these habitats (Murdoch 1998), and that these factors vary with the factor W. Direct gradient analysis can be done to determine how the coral assemblage varies relative to the factor W. As stated above, investigating the manner in which functional traits and groups of corals vary in relation to direct (environmental) gradients is most likely to produce powerful models when done along obvious physical gradients such as the one observed in the Florida Keys (Hutchinson 1957; Whittaker 1975; McGill et al 2006). All 20 sites of the Keyswide Coral Reef expedition also varied in percent cover of the total coral assemblage (hereafter referred to as “total coral cover”; Aronson and Murdoch 1996; Murdoch 1998), ranging from ~1% to ~20%. While the environmental causes of

57

Figure 2.02. Relationship between total percent cover of the entire coral assemblage and the measure of environmental disturbance due to island passes, W.

the variability of all 20 reefs is unknown, indirect gradient analysis may be used to test the hypotheses detailed in Chapter 1 regarding whether the attributes of functional groups of corals vary changes relative to total coral cover. Since total coral cover represents the most commonly-used metric for determining the ecological condition of coral reefs, and because the manner in which the attributes of functional groups of corals vary relative to total coral cover are as yet unstudied, an investigation of the manner in which functional traits and groups of corals vary in relation to the indirect (biological) gradient is also important to determine.

58

OBJECTIVES Focusing on the functional groups of corals described in Ch 1, I examined how the coral composition of 10 separate reefs located within the Middle and Lower Keys regions changed in relation to the direct environmental gradient defined by the variable W. Only these 10 reefs were surveyed relative to W as they are the only reefs within proximity to the passes connecting the reef tract to Florida Bay. In particular, I examined how the (1) rank-abundance (dominance) of species and functional groups (2) percent cover and abundance of each functional group, (3) species richness of the total coral assemblage, (4) functional group richness, and (5) species richness within each separate functional groups varied across the Middle and Lower Keys reef sites previously shown to vary in total coral cover in a linear manner relative to the distance from passes and the strength of the currents flowing through each pass (Murdoch 1998). I also investigated the relationship between the same five variables and the indirect gradient of total coral cover on all 20 of the coral reefs surveyed in the Keyswide Coral Reef Expedition, using model 2 (orthogonal) regression for functional group cover and model 1 (least squares estimate) regression for all other comparisons. While the environmental causes of the differences in total assemblage cover for the ten reefs not included in the analysis of the direct gradient were not experimentally determined, all twenty reefs were assessed for a couple of reasons. These reasons are: (1) percent cover data represent a common means of evaluating the ecological condition of coral reefs (Rogers 1994), and (2) gradient analysis of the 20 sites encompassing the entire Florida Keys region may uncover patterns in the five independent factors that are not apparent in

59

the smaller dataset from the 10 sites in the Middle and Lower Keys sectors. The 20 sites surveyed were not compared using Based on the theoretical postulates stated earlier, I predict the following patterns in the coral assemblage structure of these Floridian reefs:

Similarities in distribution of species between and among FG Species within functional groups are hypothesized to be redundant in terms of their functional responses to the environment. If this is so then species belonging to the same functional group should differ little in their distributions across sites, due to the shared responses to environmental and biological conditions (Steneck and Dethier 1994; Gitay and Noble 1997; Hooper et al. 2002). Conversely, the distribution patterns of species belonging to separate functional groups should be substantially different due to disparate environmental tolerances. Under the above paradigm, species belonging to the same functional group should differ little in their distributions across sites, due to the shared responses to environmental and biological conditions (i.e. Figure 2.03C). If, however, the traits used to distinguish species into functional groups are those primarily utilized for resource acquisition, and not environmental tolerance, then species within functional groups should be too similar in resource requirements to be able to persist within the same habitats (following Gause 1934). Under this scenario, species that belong to the same resource-based functional groups (also termed “guilds”) should tend to occur in different habitats (reviewed in Fox 1999). Also species from different guilds should be able to coexist to a greater degree than species from the same guild. Patterns of species distributions of guilds should match the pattern in Figure 2.03D, below.

60

Rank abundance by species and functional groups Examination of the rank-abundance patterns of species and functional groups was done to test whether species and functional groups differ in dominance, as predicted by the AST, or alternately, whether all coral species are functionally equivalent, as the unified neutral theory (UNT) suggests(Hubbell 2001). If the UNT is correct then all species will appear equally likely to be the most abundance on a transect, regardless of the environmental conditions of the reef in which it is found, and no species will exhibit greater dominance. Alternately, it may be that species differ in dominance, but that no within-functional group similarities in dominance patterns exist. This would indicate that each species is different from all other species, but that they do not share functional group affiliation as I previously designated them in Chapter 1. If, however, species from the same functional group tend to have similar dominance patterns across transects, this would indicate that functional groups exist and that species within functional groups tend to be fairly equivocal. In other words the UNT would apply within FG but not between FG, which is a pattern which conforms with both the AST and a version of the UNT (Hubbell 2005). If the predictions of the AST are true, then the corals of the MO (massive oviparous) functional group should dominate on reefs with high colony density. This is because they are predicted to use growth to dominate when resources are abundant and levels of disturbance low. However, since low total colony abundance was shown above to be highly correlated with W, which I showed above is a proxy for disturbance, and the MO functional group is predicted to have low growth rates and survivorship when levels of

61

disturbance or stress are high, it is predicted to not dominate on reefs with low total colony abundance. Conversely, species of the stress-tolerant and the ruderal functional groups, which are the foliose and plating functional group and the massive viviparous functional groups, are predicted to dominate reefs with low total colony density, although they should do so under different environmental conditions. The rank abundance of the species from each group will drive the rank abundance of each functional group as a whole, and if this is the case then the MO functional group is predicted to dominate most reefs with the other functional groups occupying a subordinate position. To test the null hypothesis that no species or functional group ranks higher than the rest, the counts for each possible rank was tabulated for each species, and separately for each functional group, across all 200 transects of the 20 sites. Chi-square analysis was used to determine whether the resultant table possesses a non-random distribution of ranks across sites for species and for functional groups. Logistic regression was also done to determine whether the ranks of each functional group varied in a non-random manner relative to the average coral cover of the 200 sites.

Percent cover and abundance per functional group The percent cover for the species that make up the most competitive functional group is predicted to peak on reefs far from passes, where sources of disturbance are most likely to be low and light levels likely to be highest, or on reefs with maximum levels of total coral cover. If different functional groups compete with each other then the cover of stress-tolerant and disturbance-tolerant species is predicted to peak on reefs at moderate

62

distances from passes where levels of stress and disturbance are at moderately high levels and where competitively dominant species are in relatively low abundance (Figs. 1.08A). Alternatively, if functional groups do not compete with each other to a significant degree, then the cover of all functional groups should peak in habitats within which resources are highest and sources of disturbance are low (Figs. 1.08B,C). Under this scenario, the competitive-dominant group is still predicted to have higher biomass on the reefs farthest from passes than all other functional groups. The competitive and stresstolerant functional groups of coral (i.e., MO and FP) are also predicted to have a more limited range of distribution across the disturbance gradient (W) than the disturbancetolerant functional groups of coral (i.e., Branched, Viviparous [BV] and Massive, Viviparous [MV] corals). Additionally, if the AST model is correct, the competitive dominant functional group should exhibit significantly higher percent cover on reefs with low disturbance and stress than the other groups, while the stress-tolerant and disturbance-tolerant functional groups should exhibit significantly lower percent cover, regardless of which habitats they dominate. If the predictions of the unified neutral theory (UNT) are correct all functional groups will be equivocal and will show no significant differences in distribution or in abundance or biomass proxy (i.e., percent cover) within or across reef sites. These predictions, as well as those following, were tested using both linear and second-order polynomial regressions. Linear regression were done to determine whether functional groups increased or decreased in percent across the gradients. Second-order polynomial regressions were done to determine whether the percent cover of each functional group peaked at a midpoint along the gradient.

63

Species richness of all corals Since depth was held constant among sites but turbidity most likely varied with distance from passes, the amount of resources available as light and as suspended particulate matter reaching each reef surface, as well as the level of sedimentation stress, likely varied with distance from passes as well (Shinn et al. 1989). The relative levels of stress and disturbance therefore probably vary with site location. According to the predictions of the productivity-diversity hypothesis, which lead to CSR theory (Grime 1973; 1979), and which is also known as the intermediate disturbance hypothesis (Connell 1978; Aronson and Precht 1994; Huston 1994; and many more), the habitat types with intermediate levels of stress and disturbance should display greater species richness than either the habitats with low levels of stress or disturbance and the habitats with high levels of stress or disturbance.

Functional group richness Grime (1977) predicts that the richness of functional groups will be low across sites, as each functional group is replaced by another due to the combined effects of environmental filtering and competitive exclusion between functional groups (Figure 1.08). Alternatively, if competition does not operate between FG, then all functional groups are predicted to occur on reefs characterized by the lowest values of W (or highest total cover), and functional group richness will decline as disturbance levels increase (Figure 1.08), with the distribution of less-tolerant functional groups nested within moretolerant functional groups. If the UNT is correct, then functional groups do not exist and

64

the all species have an equal chance of being found at any site regardless of position across the direct or indirect gradients.

Species richness per functional group If the predictions of Grime (1977) are correct, then the number of competitive species (i.e., those in the MO functional group) should peak on reefs in Florida that are characterized by low levels of W or high levels of total coral cover, and decline steeply as W increases or total coral cover increases. The number of ruderal and stress-tolerant species should peak at moderate levels of disturbance, due to both environmental and biological constraints. Accordingly there should be a turnover between functional groups across the direct and indirect gradients (Figure 1.08). If competition only operates between species within functional groups, however, then the species richness of all functional groups should peak at moderate levels of W or total coral cover, with the less tolerant functional group nested within the distribution of the more tolerant functional groups. This unimodal pattern would occur within FG because of the effects of competitive exclusion due to limiting similarities occurring between the functionally similar species within FG at low levels of disturbance, and environmental constraints limiting species membership within FG at high levels of disturbance (Figure 1.08).

METHODOLOGY Geographic setting

65

The Florida Reef Tract lies at the southern tip of Florida and seaward of the Florida Keys and is the third largest barrier reef in the world. The reefs and surrounding mangroves, seagrass meadows and forested islands form a large, interconnected ecosystem that sustains abundant wildlife, both above and below the sea surface (Florida Keys National Marine Sanctuary Management Plan 1995). The ecological condition of the Florida Reef Tract has been an issue of heightened national focus as coral cover and diversity have declined substantially in recent decades (Jaap et al. 1988; Murdoch and Aronson 1999; . The Florida Reef Tract is 350 km long and extends from Biscayne Bay at its northeastern most end to the Dry Tortugas islands at its western most end (Figure 2.03). A regional feature with a strong influence on the corals of the Florida reef tract is Florida Bay. Florida Bay is a large, shallow body of water with limited circulation that lies to the north and west of the Florida Keys. The water enclosed within Florida Bay is subject to extremes of temperature, salinity, turbidity, and nutrient content (Vaughan 1918, Shinn et al. 1989, Szmant & Forrester 1996). This water is forced by tides and winds onto the Florida reef tract through the tidal passes in the Middle Keys (Wang et al. 1994, Smith 1994, 1997). Reef development is poor near the passes, probably because of the effects of the water from Florida Bay on corals and other reef organisms (Shinn et al. 1989, Ginsburg and Shinn 1994).

66

Figure 2.03. Map of South Florida and the Florida Keys. The reef tract is seaward of Hawk Channel and forms a linear barrier reef roughly 350-km in length. The ten reef sites included in the direct gradient analysis are lettered E through N. All 20 of the reef sites are included in the indirect gradient analysis.

Data collection and analysis Videographic and species presence-absence data were collected during three research cruises conducted over 26 days during August, September and October, 1995. The reef tract was surveyed at twenty deeper habitats (13–19 m depth), which were chosen a priori and located from Biscayne Bay to the Dry Tortugas. Subsequent to in situ

67

assessments, sixteen of the reef sites were defined as spur-and-groove habitats and four sites as hard-ground communities. Almost all of the sites were positioned along the reef slopes of named, emergent reefs The names of these reefs were retained as designations for the survey sites (Figure 2.03). Reef ridges are frequently present offshore of the emergent reefs along the Florida Reef Tract (Lidz et al. 1991a, 1991b, 1997). These “outlier” reefs may protect the reefs that are landward of them from storm-driven waves and other oceanographic influences. The rule of the Keyswide research team was to sample the seaward-most section of reef that was at the correct depth, not necessarily corresponding to the area directly downslope from the emergent reef. For this reason, some of the reefs sampled were downslope outlier reefs. Transects were videographed and analyzed following the methodology described in Aronson et al. (1994). One or two divers stretched 25-m long waterproof tape measures down the middle of ten haphazardly selected spurs. Transects were generally 3–10 m apart and the area videographically surveyed encompassed approximately 25m x 100m of reef. Care was taken to avoid placing the transect lines over sand or off the ends of spurs. Once each transect line was in place, another diver slowly swam down its length, videotaping a 0.4 m wide x 25 m long swath of the reef. Videography was accomplished with a Hi-8 video camera that was enclosed in a underwater housing and equipped with a wide-angle lens and two 50 W waterproof lights. A 40 cm stainless steel rod projected forward from the camera housing. This rod was used as a guide so the diver could maintain a set distance of 40 cm between the camera lens and the reef surface. On the end of this rod, a 15-cm wide gray plastic bar was mounted such that it appeared in the

68

field of view of the camera. The plastic bar served as a scale in the videotaped images. During filming the underwater video camera was held perpendicular to the overall slope of the reef, with the end of the stainless-steel rod suspended less than 2 cm from the reef surface. This camera position ensured that the videographed images were of the reef in plan view. In order to accurately assess biological richness data on Caribbean coral reefs such as those in the current study, Aronson et al. (1994) determined (using species saturation curves) that more transects are required in comparison to the number of transects needed to obtain accurate percent cover data. These additional transects are needed in order to account for rare, widely dispersed species. Coral species presence data, for use in the determination of species and functional group richness per reef site, were collected in the present study from the 10 videographed transects as well as an additional 10 transects surveyed visually, following the procedure in Aronson et al. (1994). The visually assessed transects were placed in the same manner as the videographic transects, and continued over an area of roughly equal size. The richness data therefore was collected over twice the area of reef as the coral cover data. The video transects were analyzed in the laboratory with a Hi-8 videocassette recorder (VCR) attached to a high-resolution color monitor. Each transect was divided into 50 regularly spaced, non-overlapping frames, displayed by pausing the VCR. One of ten clear plastic sheets, marked with ten random points, was laid over the monitor screen and the sessile organism or substrate type present under each point identified and recorded manually. The video was then advanced to the next frame and a new sheet of dots haphazardly selected. This method is the more primitive predecessor of the computer-

69

automated method described in Chapter 4, but which results in identical data. By following this protocol, each transect yielded 500 data points. Estimated percent cover of coral species, other biological groups such as sponges and gorgonians, and of different types of substrate was calculated from the point count data from the ten transects for each reef site. The abundance of colonies for each species was also assessed from the video data. Counts of colonies were made for each species on each transect, and the results from the ten transects per site used to calculate the average abundance per species per site. The total species or functional group richness at each site was determined by summing all species present within both the ten videographic transects and ten additional transects of equal dimension that were assessed visually while diving the survey site, following the standard protocol (Aronson et al. 1994). Functional group richness was calculated using four increasingly-stringent procedures. (i) The least-stringent method allowed a functional group to be recorded as present at a site if only one colony of any member species was observed on all 20 transects. Three increasingly stringent conditions for the indication of presence of a functional group on a reef were calculated by recording member species of functional groups as present at a site when recorded over more than (ii) 5, (iii) 10, and (iv) 15 transects. Increasing the number of transects that the member species had to occur before the functional group was counted had the effect of increasingly filtering out the less ubiquitous or transient species that were present at a site.

70

Statistical analysis Similarities in distribution of species between and among FG Species that are grouped according to shared environmental tolerances should tend to co-occur in habitats, while species that share resource requirements should tend displace others from habitats. In order to examine the degree of similarity in the distribution of species species within and between functional groups, I calculated an Analysis of Similarity (ANOSIM; Table 2.01) on the raw colony count data of the 11 most abundant species across the 200 transects. Bray Curtis similarities between square-root transformed abundance data for all species were calculated and similarity trees and nonmetric multidimensional-scaled diagrams produced for visually comparison against the hypothetical patterns illustrated in Figure 1.08. To test the null hypothesis that no species or functional group ranks higher than the rest, the counts for each possible rank was tabulated for each species, and separately for each functional group, across all 200 transects of the 20 sites. Chi-square analysis was used to determine whether the resultant table possesses a non-random distribution of ranks across sites for species and for functional groups. Logistic regression was used to determine whether the rank of each functional group changed in a non-random manner relative to the total percent coral cover of each site. The predictions of the remaining hypotheses described in Chapter 1 and above were tested using regression analysis. For the tests of (1) percent cover of each functional group, (2) total species richness, (3) functional group richness and (4) species richness within functional groups against the predicted responses to both the environmental gradient W and the biological gradient in total coral cover, the correlative relationships

71

between the variables were examined by computing linear and second-order polynomial regressions. Orthogonal (Model 2) regression was used for comparing functional group cover with total coral cover, since both variables were derived from the same data, and thus had error variances of similar extent (Sokal and Rohlf 1995). Least squares estimates (Model 1) regression was done for all other variables. All second-order relationships were predicted to be concave-downward, in accordance with the hypotheses of the intermediate disturbance hypothesis (Grime 1973; Connell 1978; Huston 1994). Departures of linear regression coefficients from zero, and determinations of whether second-order coefficients for polynomial regressions were significantly different from zero, were done using one-tailed t-tests.

72

RESULTS Similarity in distribution of species within and between functional groups. Species within functional groups were hypothesized to be redundant in terms of their functional responses to the environment (Grime 1979; Steneck and Dethier 1994). In order to examine the degree of similarity in the functional responses of species within and between functional groups, I calculated an Analysis of Similarity ANOSIM; Table 2.03) on the raw colony count data of the 11 most abundant species across the 200 transects. Bray Curtis similarities between each square-root transformed abundance data for species were calculated and similarity trees and non-metric multidimensional-scaled diagrams produced for visually comparison. (Figs. 2.04, 2.05). The five of the six species of the MV functional group were more similar to each other in multidimensional distributional space than they were to corals from the other functional groups (Figure 2.05). The one exception was Stephanocoenia intercepta, which was also the least abundant member of the MO functional group. The species of the MV functional group also were more similar to each other in the patterns of abundance they exhibited across the 200 reefs than to species of the other two functional groups, with the exception of the one MO coral described above. The species of the BV functional group were dissimilar to each other and to all 9 other species in the comparison. ANOSIM analysis (Table 2..01) determined that the MV and BV functional groups were both significantly different from the MO in the manner in which the abundances were distributed across the 200 survey sites. The MV and BV functional groups were not

73

found to differ significantly in their distribution pattern, although the failure to detect significant differences may have been due to a type II error caused by the small number of possible permutations that could be run with the data. Bray-Curtis similarities were also calculated for all 36 species evaluated in the rankabundance comparisons. The many rare species appeared to display neutral patterns of distribution across sites. Visual assessment of the MDS diagram (Fig. 2.06) confirms that the rare species did not display any coherent patterns of similarity between each other. In particular, the member species of the FP group were found to have low similarity values among each other (i.e., less than 20 out of 100). The other species that displayed neutral rank-abundances across the Florida reef sites also were not similar in distribution to neither the rare nor the abundant species. ANOSIM of the entire suite of species determined that the overall pattern of abundance exhibited by the FP group was highly significantly different from that of the MO functional group (p = 0.007; Table 2.02). All other comparisons were not significant, except for the statistical comparison between the very abundant MO functional group and the very scarce BV functional group.

74

Figure 2.04. A dendrogram showing the similarities in response patterns among the eleven most abundant species assessed in Florida. Bray Curtis distances were calculated based on square-root transformed coral colony counts. Species were clustered using the group-linkage method..

75

Figure 2.05. MDS showing the similarities in response patterns among the eleven most abundant species assessed in Florida in two-dimensional state space. Species that are closer together and bound within similarity isoclines are more alike than species farther apart or outside isoclines. Bray Curtis distances were calculated based on square-root transformed coral colony counts. Species were clustered using the group-linkage method.

76

Table 2.01 Results of one-way analysis of similarity of the Bray-Curtis similarities of the abundance data of the most abundant 11 species observed across 20 reef sites located on the Florida Reef Tract.

ANOSIM Analysis of Similarities Global Test Sample statistic (Global R): 0.734 Significance level of sample statistic: 0.1% Number of permutations: 999 (Random sample from 4620) Number of permuted statistics greater than or equal to Global R: 0

Pairwise Tests R Groups

Statistic

MV, MO MV, BV 0.5 MO, BV 1

Significance Level % 0.66 10.0 0.875

Possible

Actual

Number >=

Permutations

Permutations

Observed

2.4 10 3.6

84 10 28

77

84

2

1 28

Figure 2.06. MDS showing the similarities in response patterns among all 36 species assessed in Florida in two-dimensional state space. Species that are closer together and bound within similarity isoclines are distributed across sites in a more similar manner than species farther apart or outside isoclines. Bray Curtis distances were calculated based on square-root transformed coral colony counts. Species were clustered using the group-linkage method.

78

Table 2.02. Results of one-way analysis of similarity of the Bray-Curtis similarities of the abundance data of all species observed across 20 reef sites located on the Florida Reef Tract

ANOSIM Analysis of Similarities Global Test Sample statistic (Global R): 0.073 Significance level of sample statistic: 13.5% Number of permutations: 999 (Random sample from a large number) Number of permuted statistics greater than or equal to Global R: 134

Pairwise Tests R Groups

Statistic

BO, MV

Significance Level % 0.563

Possible

Actual

Permutations Permutations

Number >= Observed

7.1

28

28

55

55

3.0

66

60.7

28

28

17

-1.000

100.0

3

3

3

MV, FP -0.033

60.7

5005

999

606

2 BO, FP

0.141

BO, MO

21.8 0.646

12 66

2 BO, BV -0.010 BO, X

79

MV, MO 0.066 MV, BV

22.8 0.022

8008

999

37.9

462

227 462

175 MV, X

-0.333

71.4

7

7

5

FP, MO

0.178

0.7

92378

999

6

FP, BV

-0.078

78.4

5005

999

783

FP, X

-0.525

100.0

10

10

10

17.3

8008

MO, BV

0.096

999

172 MO, X

-0.178

45.5

11

11

5

BV, X

-0.556

100.0

7

7

7

80

Rank Abundance per Species The six most abundant coral species out of the 34 species analyzed belonged to the massive oviparous functional group (Figure 2.07; Table 2.03). Out of the 8097 coral colonies that were counted across the 200 transects, 6330 colonies were accounted for by these six species, which represents over 78% of all the colonies observed. The probability that the six most abundant species observed would all belong to the MO functional group (out of the four groups sampled) by chance, is highly significantly small, at p = 3.61E-5. The next three most abundant species were all of the massive viviparous functional group, which were classified as ruderal corals. Represented by 831 colonies, these three species accounted for approximately 10% of all coral colonies counted. The next two most abundant species also members of a separate single functional group. These two species were both branched viviparous corals (competitive-ruderal classification), and accounted for ~4% of all colonies counted across the Florida Reef Tract, with 328 colonies total. The other 27 observed coral species accounted for less than 100 colonies each across all transects. The ranks of the most common eleven species across all transects examined individually were also distributed in a highly non-random fashion (Table 2.04; Figs 2.07, 2.08). Most transects had an average of 11 total species, of these the three top-ranking species tended to be consistently most abundant across most transects. The next three species rarely ranked below sixth on most transects. The three massive viviparous corals, which ranked seventh through ninth overall, were rarely in the top two ranks, but also rarely ranked below ninth. All species of the foliose and plating functional group ranked

81

equally as subordinate in rank, as were 4 out of 6 species of the branched viviparous functional group. A chi-square test of the distribution of all species in proportional ranks from 1 to 11 out of 34 possible rankings total was found to be very highly significantly different from the null, at p = 7.8E-192.

Figure 2.07 (below). The log percent relative abundance of the species observed at the 200 transects assessed. Different shaped points represent the different functional group memberships of the species.

82

83

Table 2.03. (below) A table of the number of occurrences with which each the 36 most abundant species ranked from 1 to 20 across all 200 transects. The designations of rank are displayed along the top row of the table. For example, the species M. faveolata ranked first on 79 transects, second on 35 transects, and so on. Darker-shaded cells represent a greater number of occurrences. A chi-squared test determined that the probability of the observed distribution of rank abundances occurred at random is exceptionally significantly small, at p = 5.22E-148.

84

85

Figure 2.08. The distribution of the proportion of ranks over the 200 sites that the most dominant species, Montastrea faveolata, displayed. The distribution of the proportion of ranks if all corals shared dominance, as predicted by the UNT, is also plotted for comparison. The null, or neutral distribution was calculated for each rank by dividing the sum of all occurrences each species by the number of occurrences of all species, and then multiplying the result by the number of occurrences for the rank in question.

86

Rank Abundance per Functional Group Since the highest-ranking 11 species were found to be most similar in dominance to other species within the same functional group, I also analyzed the rank-abundance distribution of each functional group across the 200 transects (Figure 2.09; Table 2.04). The massive oviparous functional group, classified as competitive-stress-tolerant, was found to rank highest (i.e., first) in dominance in 182 of the 200 transects. The massive viviparous functional group, classified as ruderal, ranked second compared to the other three functional groups. The BV functional group most often ranked third, and FP functional was most often fourth ranking in abundance across all sites. The ranks of each functional group were plotted against their total abundance value of each transect. This comparison was conducted to determine if the competitive - stresstolerant functional group (i.e., MO) was consistently dominate even on transects regardless of low colony abundance. Colony abundance was also shown to be highly negatively correlated with proximity to passes (Murdoch 1998) and thus is a good proxy for stress or disturbance. I also examined whether functional groups of corals hypothesized to be disturbance-or stress-tolerant would increase in dominance rank within these marginal transects (Figure 2.10). The MO functional group ranked first on reefs with more than 25 colonies. The MO group was subordinate on 18 reefs, all with low coral abundance. Conversely, the other three functional groups rarely or never ranked first in abundance except when total coral abundance was less than 25 colonies. Logistic regression of the rank data compared to total coral abundance for sites found that the patterns displayed by each functional group

87

were highly significant at p values of < 0.001. The model formulae and significance tests are presented in Appendix 2.01.

Figure 2.09. The proportion of ranks displayed by each functional group across the 200 transects surveyed. The pattern is exceptionally significantly different from a neutral distribution, at p = 1.56E-173 (Table 2).

88

Table 2.04. A table of the observed number of times each functional group ranked from 1 to 4 across the 200 transects surveyed. Also tabulated is the expected distribution of ranks under the UNT, as well as the chi-squared test comparing the differences between the two matrices. The observed pattern of rank abundances was found to be exceptional significantly different from the expected null model, at p = 1.56E-173.

Figure 2.10. (below) The relationship between rank per functional group and total abundance per transect for the 200 transects surveyed. Logistic regression analysis and tests of significant of the data are presented in Appendix 2.01.

89

90

Percent coral cover of each functional group versus W Functional groups did not appear to replace each other across the gradient of W (Figure 2.11), and instead demonstrated a nested distribution pattern as the intensity of W varied. All species peaked in cover at lowest values of W and declined in cover on reefs with higher W values. Only the MS functional group displayed high levels of cover at low levels of W, whereas all other functional groups of corals accounted for low cover across all reefs, as predicted by the CSR theory. The BV functional group did not vary significantly with W (Table 2.05; Figure 2.11a). Coral cover by the branched brooding corals was very low at all sites, and ranged from 0.09% to 1.93%. The foliose and plating, brooding functional group of corals also displayed very low percent coral cover across all sites surveyed, with a range of 0.13% to 1.51%. Despite the low cover overall, however, this functional group did display a highly significant negative correlation with W (Table 2.05; Figure 2.11b). The cover of the MV functional group of corals displayed little variation relative to W, and ranged from 0.80% to 2.40% (Figure 2.12c). The coral cover of the MV group did not vary significantly with W (Table 2.05; Figure 2.11c). Percent coral cover of the MO functional group ranged from 0.76 to 12.09% (Figure 2.12d). Coral cover for the MO group was highly-significantly correlated with W and formed a negative slope (Table 2.05; Figure 2.11d), indicating that massive corals that spawn gametes are more abundant on reefs far from island passes.

91

Figure 2.11. Relationship between percent coral cover of each functional group and environmental influence of island passes, (W). The best fit linear (–) and second-order polynomial (- -) relationships are written in the lower left-hand corner of each graph.

92

Table 2.05. Results of two-way t-test of the linear correlation between coral cover of each functional group and W on 10 reef sites along the Florida Reef Tract, testing whether the slopes are zero.

FG BO BV FP MV MO

Estimate 0.0014052 -0.009157 -0.19733 -0.018454 -0.190742

Standard Error 0.003028 0.012911 0.005124 0.010043 0.025143

t Ratio 0.4600 -0.7100 -3.8500 -1.8400 -7.5900

Prob > |t| 0.6550 0.4983 0.0049 0.1035 < 0.001

Percent cover per functional group versus total coral cover When percent cover of each functional group was regressed against total cover of all corals, only the MO functional group displayed an substantial change with cover across sites, with a range from 0.1% to 18.7% over the 20 sites surveyed (Figs. 2.12, 2.13). Alternatively, the BV, MV and FP functional groups each only displayed low coral cover (between zero and less than 2.5% cover) across the same 20 sites (Figs. 2.13, 2.14). The BV functional group displayed a significant increase in cover relative to total coral cover (Table 2.06; Figs. 2.12a, 2.13a), while the FP, MO and MV functional groups increased highly significantly with total coral cover across the 20 sites (Table 2.07, Figs 2.12b-d, 2.13b-d). Only the MV functional group displayed a significant second-order polynomial regression with total coral cover, with a concave-down regression curve (Table 2.07, Figs 2.12c, 2.13c).

93

Figure 2.12. Orthogonal relationship between average percent coral cover for each functional group and the total coral cover for the 20 reef sites surveyed on the Keyswide Coral Reef Expedition. The orthogonal fit linear (–) and second-order polynomial (- -) relationships are written in the lower lefthand corner of each graph. The thick diagonal line represents the maximum possible value of cover for each functional group relative to total coral cover.

94

Figure 2.13. The same graphs illustrating the relationship between average percent coral cover for each functional group and the total coral cover for the 20 reef sites surveyed as in Figure 2.12, but with different scales on the y-axes. The best fit linear (–) and second-order polynomial (- -) relationships are written in the lower left-hand corner of each graph. The thick diagonal line represents the maximum possible value of cover for each functional group relative to total coral cover.

95

Table 2.06. Results of orthogonal contrasts on whether the linear regressions of percent coral cover for each functional group versus total coral cover at each reef site were significantly different from zero.

FG

Variance

BV FP MV MO

Ratio 0.0077 0.0006 0.0104 0.7803

Correlation 0.4961 0.8478 0.6635 0.9894

Prob. <0.05 <0.005 < 0.005 < 0.0001

Table 2.07. Results of two-way t-tests on whether the second-order coefficients for polynomial regressions of FG cover versus total coral cover were significantly different from zero.

FG BV FP MV MO

Estimate 0.002915 0.0002887 -0.006665 0.0087727

Std Error 0.002915 0.000496 0.002531 0.004555

96

t Ratio -1.21 0.58 -2.63 1.93

Prob>|t| 0.2444 0.5683 <0.0001 0.071

Total species richness versus W A second-order polynomial regression between species richness and W was concave downward and highly significant (r2 = 0.727919, t ratio = -3.16, p = 0.016; Figure 3.14). A linear relationship between species richness and W was negative and marginally insignificant (r2 = 0.3404, t ratio = -2.03, p = 0.077; Figure 2.14).

Figure 2.14. Relationship between species richness of all corals and environmental influence of island passes, W, at each reef site. The best fit linear (–) and second-order polynomial (- -) relationships are written in the lower left-hand corner of the graph.

97

Total species richness versus total coral cover Species richness ranged from a low of 20 species to a high of 34 species out of a total number of 36 observed across all sites. A second-order polynomial regression between the total number of species per site and the total coral cover per site was concave downward and very highly significant (r2 = 0.7863, t ratio = -6.32, p < 0.001; Figure 2.15). A linear regression of the same data was also significant [r2 = 0.284602; t ratio = 2.68; p = 0.0154]. However, a plot of the residuals demonstrated that the two variables were not well represented by a linear relationship. In the plot of residuals, all but one point out of 12 between 4% and 15% cover were above the best fit linear line, and all points outside the 4% to 15% range were below the best fit line, indicating the data did not fit the assumptions of linear regression (Sokal and Rohlf 1995).

98

Figure 2.15. Relationship between species richness and total coral cover across the 20 reef sites. The best fit linear (–) and second-order polynomial (- -) relationships are written in the lower left-hand corner of each graph.

99

Functional group richness versus W Functional group richness appeared to vary depending on both the intensity of the environmental gradient W and the degree of sparseness to which species were distributed across sites (Tables 2.08, 2.09; Figure 2.16). When species observed across any one of the twenty transects per site were included, the number of functional groups observed across the 10 survey sites was 4 of 4, regardless of the value of W (Figure 2.16). However, as the constraints regarding how many transects a species must be distributed over before being counted (i.e., how ubiquitously distributed a species was) increased, a consistent pattern could be seen (Table 2.10) in which first the species within the FP functional group became increasingly sparse, then the species of the BV functional group, and finally the species of the MO functional group , Functional groups tended to lose ubiquitous species on reefs with high values of W , and so functional group richness declined as W increased.

100

Table 2.08. Results of orthogonal contrasts on the linear regressions of functional group richness for each level of constraint versus W.

Term Five Ten Fifteen

Estimate -0.011 -0.007 -0.035

Std Error 0.005 0.010 0.012

t Ratio -2.14 -0.76 -2.91

Prob>|t| 0.065 0.469 0.020

Table 2.09. Results of orthogonal contrasts on the second-order coefficients for polynomial regressions of functional group richness for each level of constraint versus W.

Term Five Ten Fifteen

Estimate -0.00059 -0.00027 -0.00083

Std Error 0.0001 0.001 0.001

101

t Ratio -3.94 -0.55 -1.59

Prob>|t| 0.006 0.600 0.160

Figure 2.16. Relationships between functional group richness under the four levels of membership constraint and the environmental gradient of W across reef sites. The best fit linear (–) and second-order polynomial (- -) relationships are written in the lower left-hand corner of each graph. The rules for inclusion of functional groups in each graph were as follows: 

ALL = At least one species per FG on one transect or more.



Five = At least one species per FG on more than five transects.



Ten – At least one species per FG on more than ten transects.



Fifteen = At least one species per FG on more than fifteen transects.

102

Table 2.10. Presence or absence matrices of the presence or absence of functional groups across the ten reefs of the environmental gradient W. The rules for inclusion of functional groups are as in Figure 2.18, above.

103

Functional group richness versus total coral cover Functional group richness declined as a function of increasing the constraints on presence in a reef site and as when total coral cover were at the highest and lowest values (Tables 2.11, 2.12; Figure 2.17). When the constraint of membership was relaxed, functional group richness was four out of four across the 20 reef sites (Figure 2.17). As the statistical constraints that filtered out sparse species within functional groups were increased functional groups decline in presence on reefs in the order of FP, then BV, the MO, while the MV functional group was observed on all reefs regardless of constraints on detection. As sparsely distributed species were filtered out of the data, FG richness took on a unimodal distribution. At highest levels of total coral cover the FP and BV functional groups were not present, while at the lowest levels of total coral cover the FP, BV and MO functional groups were absent. These results indicate that the species of the FP functional group were distributed most sparsely among transects between reefs overall, and more so on reefs characterized by high and low coral cover. The BV functional group displayed a similar pattern, but was less sparsely distributed across sites. The MO functional group only displayed absences on two reefs that had low total coral cover. The MV functional group was ubiquitous across all sites regardless of total coral cover.

104

Figure 2.17. Regression of functional group richness versus total coral cover for each site. Each graph represents a different level of membership constraint, labeled as in Figure 2.18 above. The best fit linear (–) and second-order polynomial (- -) relationships are written in the lower left-hand corner of the graph.

105

Table 2.11. Results of orthogonal contrasts on the linear regressions of functional group richness for each level of constraint versus W.

Term Five Ten Fifteen

Estimate 0.0390 0.0586 0.0567

Std Error 0.0190 0.0277 0.0305

t Ratio 2.05 2.12 1.86

Prob>|t| 0.0548 0.0485 0.0791

Table 2.12. Results of orthogonal contrasts on the second-order coefficients for polynomial regressions of functional group richness for each level of constraint versus W.

Term Five Ten Fifteen

Estimate -0.0052 -0.0143 -0.0145

Std Error 0.0029 0.0030 0.0036

106

t Ratio -1.79 -4.75 -3.98

Prob>|t| 0.0909 0.0002 0.0010

Species richness within functional groups versus W In contrast to percent cover for each functional group, in which one functional group dominated in cover, all groups demonstrated high species richness across sites relative to the strength of the environmental gradient of W (Figure 2.18). W was poorly correlated with species richness within each functional group. The BV and FP functional groups appeared to vary little in species richness except at the highest level of W. The MV and MO functional groups displayed little change in species richness across the entire environmental gradient. The species richness of the branched, viviparous, functional group ranged from 3 to 6 species per site out of a maximum of 7 species (Figure 2.18a). Species richness for the BV functional group was negatively correlated with W, but not significantly so (Table 2.13). A second-order polynomial regression of BV species richness on W was concave downward but was also not significant (Table 2.14). Species richness of FP functional group range across the 10 sites assessed, ranged from between 5 and 10 species, out of a possible 10 species observed regionally (Figure 2.18b). A linear regression of FP species richness functional group on W was not significant (Table 2.13), but a second-order polynomial regression between species richness and W was significant and concave-downwards (Table 2.14; Figure 2.18b). Neither the species richness of the MV functional group nor that of the and MO functional group were not correlated significantly with W (Table 2.13). Additionally, second-order polynomial regressions of species richness on W for both groups appeared little different from the linear regression, and were also not significant (Table 2.14, Figure 2.18c,d). The species richness of the MV functional group varied between 5 and 8

107

species (Figure 2.18c), while the MO functional group displayed a range from 8 to 10 species across the sites surveyed (Figure 2.18d).

Figure 2.18 . Relationship between species richness of each functional group and environmental influence of island passes, W, at each reef site. The best fit linear (–) and second-order polynomial (- -) relationships are written in the lower left-hand corner of each graph.

108

Table 2.13. Results of analyses of variance of the linear regression of species richness versus W for each functional group.

FG BV FP MV MO

Estimate -0.0305 -0.0508 0.0080 -0.0266

Std Error 0.0173 0.0257 0.0175 0.0145

t Ratio -1.77 -1.98 0.46 -1.84

Prob>|t| 0.1155 0.0836 0.6591 0.1034

Table 2.14. Results of two-way t-tests on whether the second-order coefficients for polynomial regressions for species richness of each functional group versus W were significantly different from zero.

FG BV FP MV MO

Estimate -0.0014 -0.0026 0.0002 -0.0003

Std Error 0.0007 0.0008 0.0008 0.0007

109

t Ratio -1.94 -3.01 0.19 -0.47

Prob>|t| 0.0939 0.0195 0.8515 0.6554

Species richness within functional groups versus total coral cover Of 36 species observed across all 20 sites, six were branched viviparous corals, ten were foliose or plating, viviparous corals, ten species were assigned to the MO functional group and nine species were characterized by a massive morphology and a viviparous reproductive mode. Across the 20 sites surveyed (Figure 2.19), the BV functional group displayed a minimum value of 2 species and a maximum of six species. The FP functional group ranged from three species to ten species. The massive oviparous functional group displayed a maximum of ten species and a minimum of seven species, while the massive oviparous functional group ranged from five to nine species across the 20 sites surveyed. The species richness of the branched viviparous functional group displayed a concave-down, highly significant second-order polynomial correlation when compared against total coral cover (Table 2.16, Figure 2.19a). A linear regression of the same data was not significant (Table 2.15). A second-order polynomial regression of the species richness of the FP functional group was also highly significantly correlated with total coral cover, and also displayed a concave-down curve (Table 2.16, Figure 2.19b). When the FP functional group data was divided between sites with coral cover values less than 10% and values greater than 10% in a post-hoc analysis, it was subsequently found that species richness of the FP group was linearly negatively correlated with total coral cover at a very high significance level (Table 2.15, Figure 2.19b). In contrast, the species richness of the FP group was not correlated with total coral cover at sites with coral cover values great than 10% (Table 2.16, Figure 2.19b).

110

In comparison with the BV and FP functional groups, the species richness of the MO functional group remained at roughly 8 species across all sites, with no significant change regardless of total coral cover (Table 2.15, Figure 2.19c). The species richness of the massive viviparous functional group also remained high across all reefs and was found to slightly increase in relation with total coral cover. However, linear regression analysis determined that the MO functional group exhibited a marginally insignificant correlation with total coral cover (Table 2.16, Figure 2.19d). The second-order polynomial regression of the species richness for both the MO and MV functional groups versus total coral cover per site was not significant (Table 2.16).

111

Figure 2.19. Regressions of species richness for each functional group on total coral cover for each site. The best fit linear (–) and second-order polynomial (- -) relationships are written in the lower left-hand corner of each graph.

112

Table 2.15. Results of two-way t-tests of the linear regression of species richness for each functional group versus total coral cover for each functional group.

FG BV FPa FPb MV MO

Estimate 0.0336813 0.6828664 -0.001081 -0.02411 0.0590496

Std Error 0.052473 0.111125 0.050731 0.035353 0.030219

t Ratio 0.64 6.15 -0.02 -0.68 1.95

Prob>|t| 0.5290 <.0001 0.9838 0.5039 0.0664

Table 2.16. Results of two-way t-tests on whether the second-order coefficients for polynomial regressions of FG species richness versus total coral assemblage cover were significantly different from zero.

FG BV FP MV MO

Estimate -0.02798 -0.032122 -0.008554 -0.001785

Std Error 0.005511 0.006295 0.005512 0.005016

113

t Ratio -5.08 -5.1 -1.55 -0.36

Prob>|t| <.0001 <.0001 0.1391 0.7263

Analysis of the presence or absence of each species across all sites in a sorted matrix displayed the manner in which the species of each functional group responded to the environmental differences that occurred across the 20 reef sites (Table 2.17). All species included in the BV functional group were present at sites possessing moderate total coral cover, but that functional group lost species on reefs that had the highest and lowest percent coral cover values. The species Madracis decactis and Porites furcata were present on all sites, whereas Porites divaricata and Madracis mirabilis were absent from reefs with both exceptionally high and exceptionally low coral cover. The FP functional group displayed a linear decline in species richness on reefs with less than 10% coral cover. Visual analysis of a sorted matrix of the FP functional group data illustrates this decline. All of the 10 FP species are present on at least some of the reefs possessing 10% cover or more, but species are lost progressively across sites with lower cover values. Only Eusmilia fastigata was found at all 20 sites, whereas Mussa angulosa, Mycetophylia lamarkiana, My. ferox, My. danaana were absent from virtually all reefs with less than 5% cover. In contrast to the BV and FP functional groups, the species of the MO functional group are present across all sites regardless of coral cover according to visual examination of a sorted matrix of this group. Only the shallow-water species Montastraea annularis is rare across all sites. Examination of a sorted data matrix of the MV functional group illustrates that the species of this group vary little in presence or absence across all sites regardless of total coral cover. One notable exception is Favia fragum, which is present at sites with a total

114

coral cover below 10%, but not present at sites with a total coral cover above 10%. Isophllastrea rigida and Isophyllia sinuosa are absent from almost all sites.

Table 2.17. (below) Sorted matrices of species presence or absence for each functional group across 20 transects at each of the 20 sites of the Keyswide Coral Reef Expedition. White cells represent absent species, grey cells represent present species. The heavy black line running across each matrix represents the predicted boundary between present and absent species if the perfect nestedness of sorted species had occurred at each reef site. Species absent below the bordered cells are expected to be present, and species present above the bordered cells are predicted to be absent. Notice that the pattern of bordered cells across sites in the table matches the pattern of points in the graphs of species richness per functional group versus total coral cover (Figure 2.19). MDP values represent the mid-point location of each species across the gradient of reef sites, calculated by reciprocal averaging of the sites of occurrence of each species.

115

116

DISCUSSION Functional groups of corals were observed to respond to direct and indirect gradients of disturbance in a manner that substantiates the premise that each group of species allocate different functional responses to environmental and biological conditions on a reef. Despite the chaotic patterns in biomass displayed by each coral species when separately plotted across reefs (Figure 1.01), each functional group of corals responded to the direct and indirect gradients in a orderly and group-specific manner. The gradient replacement pattern predicted by Grime (1977), in which each functional group dominated a particular region of the gradients, was not observed. Instead, functional groups displayed a nested distribution pattern, indicating that if negative interactions occur between functional groups they have little impact on distribution patterns. Instead coexistence strategies of some form may be operating between functional groups or functional groups have such different life-history strategies that they rarely interact. The ecological significance of each statistical test is discussed below.

Dominance by species and functional groups Examination of patterns of rank abundance determined that on the reefs considered in this analysis, eleven species out of a pool of at least 36 species were substantially more abundant than the rest (Figure 2.05). These species belonged to three different functional groups and the species members of each group both (1) sorted by rank in a similar manner and were (2) distributed across the 200 transects surveyed in a statistically similar way across multivariate state space (Figs 2.10, 2.11, 2.12). The rarer species, alternately, showed rank-abundances (Figure 2.05) and multivariate distributions across transects

117

(Figure 2.12) that were without apparent pattern. These results indicate that all coral species are not equivalent in their functional responses. These results alone refute one of the fundamental hypotheses of the UNT (Hubbell 1997, 2001) that all species of coral are functionally identical and can be modeled accordingly. By ignoring the shape of the rank-abundance curve, which is statistically similar to many other curves (McGill 2006), I instead focused on manner in which each species was ranked across transects, while taking into account its functional characteristics and the manner in which each species and functional group was predicted to display abundance patterns across gradients of disturbance and stress. This species-specific and functional-group specific analysis uncovered exceptionally non-neutral distributions of rank by the 11 dominant species within the habitat. Furthermore, species of corals that shared life-history and other characteristics were distributed across transects in statistically similar patterns to other members of the same functional group, confirming that functional groups are composed of functionally (and statistically) similar species that all differ substantially in functional responses when compared with the responses of the member species of other functional groups. The functional-group differences in rank-abundance were also highly non-random (Table 2.02; Figs 2.08, 2.09). Furthermore, the relative levels of dominance by each functional group in relationship to the total colony abundance of the set of 200 transects fit the predictions of the AST, based on the life-history strategies that each functional group possesses. For instance, the functional group that was predicted to be the competitive dominate in the Florida reef habitat was the massive, oviparous functional group, which is made up of corals with a domal morphology, generally large colony size

118

and a seasonal reproductive pattern. Theoretically another functional group, the branched oviparous corals such as Acropora cervicornis would be considered more likely to dominate reefs in good condition in Florida, but the members of this functional group were ecologically removed by disease and coral bleaching (Aronson and Precht 1998). With the BV functional group no longer ecologically important on the Florida reefs the MO functional group are the fastest growing as adults and the most capable of allocating resources for growth and domination of benign reef environments. The MO functional group consistently ranked highest of the four functional groups across each of the 200 transects assessed.(Figure 2.08), except when the total colony abundance of a transect was below 25 colonies per transect (Figure 2.09). These transects with low total colony abundance were shown in Murdoch (1998) to represent heavily disturbed areas of reef. High levels of disturbance would lead to colony damage that would likely outpace the ability of the MO corals to acquire new resources for repair. The more weedy or more stress-tolerant corals of the BV, MV and FP functional groups are theoretically better suited to habitats exposed to higher levels of stress or disturbance, and the results above did determine that these three groups each ranked highest in dominance only on the transects with total colony abundance below 25 colonies per transect (Figure 2.09). Other patterns of rank abundance also matched the predictions of the AST. The FP functional group is characterized as stress-tolerant in the functional group framework I developed in Chapter 1. Stress-tolerant species are predicted to have slow growth rates, low rates of reproduction and to rarely fragment. For these reasons, colony abundance of the FP functional group was expected to be ranked lower than that of the other functional groups, a priori. Examination of the rank-abundances of each species and of the FP

119

functional group as a whole both confirmed that this functional group was consistently lowest ranking in abundance on the vast majority of transects. Unlike the FP functional group, the species of the MV functional group have a ruderal life-history. These corals primarily utilize resources for reproduction and have higher levels of recruitment than other types of corals. Although they also are poor competitors for space, high levels of recruitment may permit the MV functional group to maintain a secondary dominance on most reefs of high total colony abundance. On reefs with low colony abundance the MV corals would be freed from competition with the MO functional group, which may also allow the MV functional group to dominate these habitats.

Percent coral cover per functional group Grime (1977) predicted that different functional groups of sessile organisms would dominate (in terms of biomass) specific zones of a disturbance gradient, due to: (1) their distinct environmental tolerances limiting membership as disturbance levels increase across sites; and (2) negative biological interactions between groups inhibiting membership as disturbance decreases across sites (Figure 2.03a). Alternatively, if functional groups are primarily affected by environmental conditions alone and do not compete with each other, then an alternative pattern was hypothesized, in which the ranges of the least environmentally-tolerant functional groups are nested within the ranges across the environmental gradient of the more tolerant functional group. Linear and second-order polynomial regression analysis of functional group percent cover (used as a proxy for biomass by coral ecologists) and functional group richness across both a

120

direct environmental gradient and an indirect biological gradient demonstrated that, on the deeper spur-and groove fore reefs in the Florida Keys, coral functional groups conform to the nested pattern with high overlap (i.e., Figure 2.03c) and not the replacement pattern (Figure 1.03d). The MO functional group, which is composed of large, broadcast-spawning head corals, exhibited the highest cover at the least disturbed site in the direct gradient and indirect gradient analysis. Additionally the MO functional group was consistently the dominant functional group across all sites except at the individual sites where the environmental measure of disturbance (W) was highest, or total coral cover was lowest. Since the branched, oviparous corals are ecologically extinct from the Florida Keys region (Precht and Aronson 2004), the massive oviparous corals represent the most competitive coral group remaining. As such, the dominance of the MO group fits the predictions of the AST model (Grime 1979). Other functional groups of corals were predicted by the AST (Grime 1979) to peak in biomass at some moderate position across the direct and indirect gradients. However, only the MB functional group displayed a significant (p < 0.001) unimodal distribution, which was observed in the indirect gradient analysis. The BV functional group did also exhibit higher values in cover at moderate total coral cover and a downward facing second-order polynomial relationship, but the polynomial section of the equation was not significantly different from zero (Table 2.05). The best fit second-order polynomial relationship of the FP and MO functional groups did not differ visually from the best fit line of the linear regression for each functional groups. Additionally, while all functional groups displayed significant linear relationships with total coral cover, only the FP and

121

MO functional groups displayed a significant linear relationship with W. As such, the distribution of the MO functional group across the direct and indirect gradients, and its biomass dominance across reefs, fits the predictions of the AST model (Grime 1979). The Ruderal and Competitive Ruderal functional groups of corals were predicted by the AST (Grime 1979) to peak in biomass at some moderate position across the direct and indirect gradients. For Caribbean corals, these adaptive strategies are represented by the massive, viviparous corals and the branching, viviparous corals, respectively. The MB functional group did display a significant (p < 0.001) unimodal distribution in percent coral cover across the indirect gradient analysis. The BV functional group did also exhibit higher values in cover at moderate total coral cover and a downward facing second-order polynomial relationship, but the polynomial section of the equation was not significantly different from zero (Table 2.05). Another prediction of the AST (Grime 1979) was that all functional groups will display lowest percent cover at highest levels of disturbance. This prediction was also observed in the analysis of the direct gradient (Figure 2.04; Table 2.01).

Total species richness According to both the IDH (Connell 1978; Huston 1994) and the AST (Grime 1979) a peak in species richness is expected to appear at a location across either direct or indirect gradients where rates of population growth and rates of recovery from disturbance are both at moderate levels (Figure 2.16; 2.17). Total species richness was found to have a highly significant unimodal relationship with W, and a very highly significant unimodal relationship with total coral cover. Linear regression across the direct environmental

122

gradient was not significantly different from a flat line. For species richness data measured across a gradient of coral cover, examination of the residual plot determined that linear regression, although significant, was not an appropriate model. It is interesting to note that even at very low values of total coral cover, the species richness of a site was greater than 20 species, which could be considered very high for the Florida region at this depth, (Dustan and Halas 1987; Porter and Meier 1992; Aronson et al 1994). This high species richness despite low coral cover is probably due to the inclusion of rare and sparsely distributed species over the relatively large area (~200 m2) surveyed for species at each site. It is worth noting here that the IDH as a theoretical model is specifically intended to explain species richness and diversity at the patch scale, and that at regional scales a linear relationship between biomass and species richness would be expected (Chase and Leibold 2002). The results presented here demonstrate that the IDH also applies at the meso-scale on the Florida Reef Tract.( i.e., between reefs separated by ~ 10 km across the 350-km long region), despite the large area encompassed.

Functional group richness. Species belonging to all four functional groups were found on every reef site surveyed in Florida, regardless of the position of reefs along either the direct or indirect gradients of disturbance. As such, coral assemblage structure did not match neither the Gradient Replacement pattern predicted by Grime (1977) nor the Nestedness pattern predicted by Patterson (1987) at the level of functional groups. The lack of turnover of functional groups across the reef sites indicates that at least one species within each functional group is capable of surviving on all of the reefs surveyed. Sites that were not spur-and-groove

123

reef were not considered in the analysis. The exclusion of marginal reef habitat no doubt limited the degree of extreme environmental conditions encompassed by the gradient analysis. Additional research that includes reef and non-reef habitats would be required to examine whether environmental conditions can limit functional group presence on the Florida Keys reefs.

Species richness within functional groups Some species displayed a limited range of reefs across which they were present within all functional groups. Meanwhile, other species within both functional groups were present across all sites in both gradients. The nested pattern was most strongly displayed by species within the FP functional group, with highest richness found in low disturbance reefs and a linear decline in richness as total coral cover declined. The pattern exhibited by the FP species was expected, as the FP functional group possesses a stress-tolerant adaptive strategy that was predicted to be most affected by increased levels of disturbance. The species of the BV functional group also displayed a nested distribution pattern, except in this case the peak in richness was found at a mid-point on the indirect gradient and the same species were absent from both ends of the gradient. This pattern indicated that some species within the BV functional group are tolerant of a wide range of environmental conditions while others are not. Alternately it could be that environmental conditions limited recruitment or colony survival at one end of the gradient and competition or predation limited membership at the other end of the gradient (Menge and Sutherland 1987). Examination of recruitment densities, partial mortality, competitive

124

interactions and indicators of predation, such as fish bites, would be required to determine the cause of the unimodal pattern exhibited by the BV functional group.

Some functional groups displayed significant changes in species richness across the indirect gradient of total coral cover across the 20 reef sites, however. Both the BV and FP functional groups displayed significantly fewer species on reefs characterized by low coral cover. The foliose, plating and solitary corals of the FP functional group possess many traits that promote the tolerance of stress, which Grime (1977) predicted would restrict their ability to persist under conditions of high disturbance. Examination of the sorted matrix of FP species shows that the four species most affected negatively by the indirect gradient are Mycetophyllia lamarikiana, Mussa angulos, My. ferox and My. danaana. These corals all possess larger polyps, thicker tissue and are the most aggressive of the FP functional group (Budd et al. 2001; Lang 1972). The sensitivity of the FP species, especially species from the Mycetophyllia genus, to disturbance indicates that these K-selected and typically rare species may be most susceptible to degraded water quality from Florida Bay. The branched, viviparous corals that were excluded from reefs with low coral cover include Porites divaricata, Madracis mirabilis and Madracis formosa. The same species were also absent on reefs with the highest levels of total coral cover. These corals possessed the smallest polyps and thinnest branches of the species within the BV functional group (Budd et al. 2001). This pattern indicated that some species within the BV functional group are tolerant of a wide range of environmental conditions while others are not. Alternately it could be that environmental conditions limited recruitment

125

or colony survival at one end of the gradient and competition or predation limited membership at the other end of the gradient (Menge and Sutherland 1987). Examination of recruitment densities, partial mortality, competitive interactions and indicators of predation, such as fish bites, would be required to determine the cause of the unimodal pattern exhibited by the BV functional group. Both the MV and MO functional groups displayed very little change in species richness across reefs, regardless of total coral cover. It was predicted that corals of the MV functional group would be more tolerant of disturbance than species from other functional groups, since the MV corals exhibit high levels of recruitment and a smaller overall colony size, as described in Ch. 1, above. One or two species from the MO and MV group did display a reduced range of sites across which they were found. In the MV group Montastrea annularis was only found at four reef sites out of the 20 included in the analysis of the indirect gradient. M. annularis is known to dominate shallow water habitats. It is probable that the limited distribution displayed by this species on the deeper forereef was due to these sites being outside the range of its environmental tolerance. Three species were found to have a limited distribution across the indirect gradient of 20 reefs. Isophyllastrea rigida was only found at one site, while Isophylia sinuosa was observed at six reef sites. The rarity of both species implies that the environmental conditions of the deeper forereef sites are also outside the tolerances that they are adapted too. The response of the MV coral Favia fragum is particularly worthy of note; it was only found in sites with low total coral cover and not in sites with high total cover, and had the lowest midpoint value of all corals regardless of functional group membership

126

(Table 2.08). It may be that F. fragum selectively recruits to habitats characterized by high disturbance, or that it is particularly susceptible to competitive exclusion by more aggressive species. The species richness of the MO functional group did decline as total coral cover decreased, but the relationship was marginally insignificant. Perhaps the large scale at which the surveys were taken allowed for the inclusion of less-disturbed microhabitats which allowed for the persistence of this functional group. Alternately, the classification of the MO functional group as Competitive–Stress-tolerant may need to be revised. However, since coral cover of this functional group greatly exceeded that of all other groups, it does conform to the competitive dominant strategy as proposed by Grime (1977) in other ways.

CONCLUSIONS In the introductory chapter I described how, when graphed, the percent cover data for all corals surveyed (redrawn as Figure 2.20 below) on the Keyswide Coral Reef Expedition produced a chaotic pattern in which each species appears to vary independently. I used the confusing graph to demonstrate the need for techniques with which to organize and simplify species-level data so that they can be interpreted in an ecologically meaningful manner.

127

Figure 2.20. Percent cover for each of the 38 species recorded on the 20 reefs surveyed on the Keyswide Coral Reef Expedition.

The functional group framework that I developed in Chapter 1 was tested in this chapter using data collected from the same corals as in the above example. described in Chapter 1. As such we should now be able to re-organize the species data that produced the chaotic graph above into functional groups, and expect to see meaningful pattern where, in the figure above, there is none.

128

Figure 2.21. Functional group cover for each of the four predominant functional groups recorded on the 20 reefs surveyed on the Keyswide Coral Reef Expedition.

As can be seen in Figure 2.21, the functional group framework provides a means for interpreting the species data in a more ecologically meaning way. The MO functional group, which dominates reefs with low levels of disturbance, can be seen to peak on reefs opposite the islands of the Florida Keys. On reefs near passes through the islands, however, the percent cover of the MO functional group declines dramatically. The other three functional groups are not predicted to be as abundant or to produce as much biomass as the MO functional group, and we can see that they remain at low overall cover even on the reefs on which the MO functional group can accumulate high levels of biomass. Figure 2.21 also indicates that since the species of the MO functional

129

group are the predominant provider of coral cover in the Florida keys, management action should focus on protecting these species or promoting their recovery. The other functional groups are less affected by disturbance and stress than the MO functional group, however, and it may be advantageous to determine what life-history strategies allow these more tolerant corals to persist in the face of stress or disturbance so that all coral species may be better protected from environmental change.

130

131

CHAPTER 3. THE GEOGRAPHY AND ENVIRONMENTAL CHARACTERISTICS OF THE NORTH LAGOON OF BERMUDA

INTRODUCTION Bermuda, a United Kingdom Overseas Territory, is located on a 5560-hectare chain of limestone islands located in the North Atlantic near 32oN 64oW (Figure 3.01). Although Bermuda is north of the tropics, prevailing warm oceanic conditions support a limited number of small mangrove forests, extensive seagrass beds and well-developed coral reefs. The Bermuda reef platform encompasses a wide range of habitat types, from small, enclosed bays and harbors to the broad lagoon, all of which are encircled by a welldeveloped rim reef and large, exposed fore-reef zones. Bermuda is host to a reduced suite of species relative to more southern reefs of the Caribbean, with only 22 species of shallow-water hard coral recorded (Appendix 2.01; Sterrer 1998). The relatively small number of local species aids in the search for biota - environment linkages, by reducing the complexity of the coral assemblage. Additionally, there exists a library of climatic and oceanographic data on Bermuda’s marine environment that dates back many decades (Bermuda Weather Service; Bermuda Institute of Ocean Sciences [formerly the Bermuda Biological Station for Research, Inc.]).

132

It appears that acroporid corals were not present in Bermuda over the past several hundred thousand years (Garrett et al. 1971). Repeated transplant experiments carried out in the early 1970s at a site on the northern rim reef confirmed that acroporid corals are currently prevented from establishment on Bermuda reefs by cold winter temperatures (R. N. Ginsburg and E. A. Shinn personal communication). Consequently, unlike most of

Figure 3.01. A photomosaic map of the Bermuda Islands and surrounding reef platform.

133

the western Atlantic (Aronson and Precht 2001), Bermuda’s reefs were not affected by the loss of staghorn and elkhorn corals that occurred in the 1980s to early 1990s. Instead, in most places the reef community appears to have changed little in the past 30 years when compared to the rest of the Caribbean, despite continued perturbation from overfishing (Butler et al. 1993), ship groundings (Smith 1992), sedimentation (Dodge and Vaisnys 1977), land reclamation (Flood et al. 2006), coral bleaching and coral disease (Cook et al. 1990). Coral cover in Bermuda averages 50–90% on the terrace reef (Logan 1988), 20–26% at rim reef sites (Dodge et al. 1982, CARICOMP 1997; Murdoch et al. unpublished data), 17% (with a range of 10–45%) on patch reefs (Dodge et al. 1982, Garrett et al. 1971; Murdoch et al. unpublished data) and 13% inside the breaker line on the South shore (Garrett et al. 1971; Murdoch et al. unpublished data).

Study Area The area under study is centrally located within the north lagoon, and is bounded at the southern extent by the north shore of the main island of Bermuda (Figure 3.02). Roughly 1-km seaward of the shore runs the South Ships Channel, which is ~100-m wide, 10–13 m deep, and which starts seaward of the island of St. Georges, at the east end, and continues westward to Grassy Bay, where it branches and continues to three locations. One branch of the channel extends to connect to the Royal Naval Dockyard, while the other branch continues through the Great Sound and there splits again, with one secondary channel continuing on to the anchorages on either side of Morgan’s Point and the other secondary channel turning eastward and terminating at the docks in Hamilton Harbour.

134

Platform margin (fore) reefs and lagoonal reefs have been historically classified into several different types, with some of the nomenclature specific to Bermuda (Figure 3.02; Appendix A; Garrett et al. 1971; Logan 1988; Logan and Thomas 1992). Within the lagoon, pinnacle reefs are characteristically steep-sided patch reefs measuring 10–150 m in width and 6–20 m in height, and are typically found in the outer lagoon. Ring-shaped patch reefs that are 50-m to 500-m wide are known as “mini-atolls”. Mini-atolls typically have a raised rim of coral and algae encircling a sediment-filled mini-lagoon containing only scattered coral knobs. When mini-atolls extend beyond 500 m in width they are referred to as “faro” reefs (Stoddart and Scoffin 1979; Logan 1988). In Bermuda faro reefs generally exhibit large central areas of shallow (~4-m depth) sandy seabed peppered with very sparse coral knobs, fringed by a ~10-m wide ridge of well-developed coral reef and surrounded by much deeper water (10 - 20-m depth).

135

Figure 3.02. (below) An illustrated map of the islands and surrounding lagoonal patch reefs of Bermuda, with important geographic features labeled (produced by the author). The islands of Bermuda are clustered in a fishhook-like shape on the southeast side of the atoll. The reef platform extends 15 km to the northwest of the island. The rim reef reaches to within 2 m of the sea surface and encloses the north lagoon and the tens of thousands of patch and pinnacle reefs therein. The fore reef surrounds the island and platform and extends down to a continuous field of loose rubble rhodoliths that rings the platform at roughly 100-m depth. Below this field, the sides of the extinct volcano on which Bermuda rests continue down without coral growth to the Bermuda Rise, over 4000-m deep.

136

137

The study area is characterized by several large clusters of patch, pinnacle and faro reefs that extend across the north lagoon (Figure 3.02). Baileys Flat (or Flats), which form the eastern boundary of the study area, are a string of well-developed patch reefs that runs in a roughly linear fashion for 9 km, from just offshore of Bailey’s Bay out to a large shallow faro reef called The Crescent. A similar congregation of patch reefs, known as Brackish Pond Flats, is found near the western edge of the study area. Brackish Pond Flats extend in a north-south direction from Spanish Point out towards Devils Flats. Brackish Pond Flats lies about 5 km west of Bailey’s Flats. Both of these linear reef systems are thought to have formed on the tops of topographic highs that were once aeolian dunes (windblown sand) formed during glacial periods, when sea levels were up to 100-m lower than at present and the Bermuda platform was subaerially exposed (Garrett and Scoffin 1977). Connecting the northern end of the two reef flats is a network of faro reefs called White Flats and The Crescent. These reefs possibly grew on top of massive wash-over sand deposits that were created by the storm erosion of sand islands that are hypothesized to have been in place along the northeast rim during the last rise in sea level, roughly 6000 years ago (Garrett and Scoffin 1977). Within the basin bound by Baileys Flats, Brackish Pond Flats and White Flats are found scattered patch reefs. These patch reefs tend to reach to very close to the surface and are generally conical in shape. Seaward of White Flats and The Crescent lies a natural channel that is ~20-m deep, and through which runs the North Ships Channel. Northward of the shipping channel are located several lobe-shaped faro-reef formations, 1–2 km in length and up to 1-km in width, and which also are hypothesized to have formed on top of wash-over deposits

138

(Garrett et al, 1971). There is currently >20 m of coral accretion on top of the Pleistocene deposits, which grew after submersion 6-8 thousand years ago (ka) (Garrett et al, 1971; Garrett and Scoffin 1977). Many flat-topped pinnacle reefs, each roughly 50–100 m in diameter, are scattered between the faros in this area. The faro reefs extend in a northwesterly direction to adjoin at the landward side of a section of the rim reef locally known as the Ledge Flats. Seaward of the rim lies the reef terrace, forereef, and then the volcanic slope of the Bermuda seamount which continues down to the abyssal depths. Eastward of Baileys Flat is a large expanse of flat sandy seabed at 13 m depth and with no coral reefs, known as Murray’s Anchorage. Westward of Brackish Pond Flats is the entrance to the Great Sound, and the western side of the North Shipping Channel, as well as Dockyard. Further to the west continues the lagoon, across which are scattered other reef flats and clusters of patch reefs. All surveys were carried out within the study area extending from land out to the rim reef across the middle of the North Lagoon. The study area was partitioned into six zones representing different distances from shore (Figure 3.07). Sites were not surveyed on the rim or fore reef as these areas are an ecologically and hydrographically a different kind of habitat. The forereef habitat is exposed to oceanic water over at least half of the tidal cycle and receives the full brunt of storm waves. In contrast, lagoonal waters are retained on the platform for 4.2 days and have higher loads of suspended sediment, a greater range of temperature and lower levels of nitrogen than offshore oceanic waters (see Morris et al. 1977). In addition, lagoonal reefs are protected from the full strength of oceanic waves emanating from the N, E and W quarters by the encircling rim reef, and from waves originating from the S by the islands of Bermuda. Furthermore, the rim and fore-reef

139

habitats are all interconnected and form one large ring-shaped reef, while the patch and faro reefs are often separated from neighbors by expanses of relatively deep water and sediment-covered basins. Below I review the previously published information regarding the distributions of marine environmental factors or reef corals across habitats in Bermuda. Subsequently I describe three separate projects in which I quantified different aspects of the study area for which information was lacking previously. In Project 1, I describe how the coral reefs located within the study area was mapped into a Geographic Information System application (Project 1). In Project 2 I describe the results of an effort to quantify light availability across the study area. Project 3 details the results of an analysis that determined how light varied in intensity over a range of depths and when measured originating from different directions.

Previous research into the distribution of corals across the North Lagoon The coral reefs of Bermuda have been the focus of interest for geologists and biologists for over 100 years (Heilprin 1889; Agassiz 1895; Verrill 1902). Recent research has focused on reefs within Castle Harbour (Dodge et al. 1982; Smith 1999; Flood et al. 2006; Jones et al. 2007), along the nearshore zones off North Shore (Smith et al. 1998; Jones et al. 2007), within the central lagoon at Three-Hill Shoals and Crescent Reef (e.g. Logan 1988; Smith et al. 2003; Jones et al. 2007), and on the northern and southern forereef terrace (Cook et al 1996; Jones et al. 2007). These previous surveys found that coral cover across the study area of the current project increases with distance from shore, with reefs near North Shore having the lowest cover (10-15 percent coral

140

cover; Figure 3.02), lagoonal reefs at Crescent and Three Hill Shoals having higher cover (25-35 percent coral cover), and forereefs in the area around North Rock having the highest coral cover (35-45 percent coral cover; Jones et al. 2007). Nearshore sites were hypothesized to have lower cover due to the water quality of the area limiting growth and survivorship. Water quality was hypothesized to be poor nearshore due to the proximity to the reefs to areas of high population density and to the southern shipping channel, which is a source of increased sediment suspension (Jones et al. 2007). However, prior to the research project described in this dissertation, no data existed for the patch reefs at intermediate locations between North Shore and the Crescent and Three-Hill Shoals area. Also, only two lagoonal patch reefs in the zone just shoreward of the platform rim have ever been assessed. These two reefs were surveyed in the early 1970s (Garrett et al. 1971; Logan 1988), before overfishing drastically reduced the densities of sharks and other carnivorous fish species such as rockfish (i.e. groupers) and snappers (Butler et al. 1993). In Chapter 4, I investigate whether coral cover, coral richness, and the relative contribution by different functional groups of corals do vary with distance from shore in a linear fashion, as the previous research described above indicates, or whether a different pattern really exists. I also compare sites that differ in depth and aspect across patch reefs, which represents another level of analysis that was not attempted by prior researchers.

141

Predominant environmental factors in operation across the study area The environmental factors generally considered to most strongly determine the assemblage structure of reef corals in Bermuda are temperature (Cook et al. 1990; Smith 1999), light (Dodge and Vaisnys 1977; Fricke and Meischner 1985; Logan et al 1994), wave energy (Upchurch 1970; Mills et al. 2004), and the amount of suspended sediments in the water column (Smith 1999; Mills 2000; Mills et al. 2004). These four factors interact over environmental gradients that occur across the area of study (Logan 1988).

Suspended Particulate Matter High levels of suspended sediment, or particulate matter (SPM), which reduces light transmission to depth, are generally found in nearshore areas and decrease in concentration as one moves offshore (Upchurch 1970; Logan 1988; G. Toro Farmer unpublished data). This gradient is most likely present because the nearshore reefs are protected from the predominantly southwesterly winds over the year by the location of the island along the southern side of the reef platform. The sheltering effect of land reduces the yearly amount of wave energy, allowing the retention of finer-grained sediments near the shore (Upchurch 1970). The nearshore waters of the north lagoon also experience a dramatic increase in the traffic of ships through the south channel in the summer (Figure 3.03). These ships typically re-suspend substantial amounts of sediment, as seen in the long, wide sediment trails that remain along the entire length of the ships channel for many hours (Figure 3.04). The southern shipping channel runs along the north shore of the island at a distance of roughly 0.5 km from shore (Figure 3.06, below).

142

The lagoonal reefs far from shore are not as well protected by the island from waves generated by high wind, although neighboring shallow (~2-m depth) reefs probably reduce wave energy to some degree. Also, while there is a northern shipping channel that passes through the offshore patch reefs, during the sampling period of this project (2000– 2005) ships only rarely traversed this passage (Bermuda Government, Dept. Marine and Ports). For these reasons the reefs offshore experience higher wave energy (Mills et al. 2004), but also lower levels of suspended sediment. As I show below (Project 3), the lower levels of suspended sediment offshore allows more light to reach a given depth on offshore reefs than at a comparable depth on inshore reefs.

143

Figure 3.03. A graph charting the number of passages by ships traveling through the southern shipping channel in 2004. Cruise ships carrying tourists only come to Bermuda in the warm months of summer. Ships bringing cargo visit Bermuda on a regular weekly schedule over the entire year. The data presented is derived from the Bermuda 2004 Cruise Ship Schedule (Bermuda Govt., Department of Tourism and Transport 2004).

144

PLUME

Figure 3.04. An aerial photograph of a cruise ship traversing the south shipping channel of Bermuda in a westerly direction, and leaving a plume of sediment in its wake. A portion of North Shore and Spanish Point can be seen behind the ship. (© 2006 T.J.T.Murdoch)

Water Temperature Temperature varies in a predictable gradient from nearshore to offshore across the study area (CARICOMP 1997; de Putron 2003). Reefs near the rim are exposed to oceanic waters, which exhibit a limited range in temperature relative to the seawater in enclosed bays and harbors. For this reason the greatest range in temperature occurs in inshore waters, which cool to below 17°C in the winter months (February-April) and warm to 31°C in the summer months of August and September (Figure 3.05). Offshore reefs, alternatively , reach 18 to 19°C in winter and 29°C in summer months. Outer

145

lagoon reefs experience temperatures that lie between these two extremes (de Putron 2003).

29 26 23 20

Nov-00

Sep-00

Jul-00

May-00

Mar-00

Jan-00

Nov-99

Sep-99

Jul-99

May-99

Mar-99

Jan-99

Nov-98

Offshore reefs

Sep-98

Inshore reefs

Jul-98

Jan-98

14

May-98

17

Mar-98

Temperature °C

32

Figure 3.05. Temperature record for 2 subsurface temperature data loggers located either near North Shore (Inshore) or on the forereef at 30 ft depth (Offshore) for a three year period from 1998 – 2000 (modified from de Putron 2003).

Solar Radiation The Effect of Depth and Turbidity The visible sunlight that we see is of the same general range of wavelengths as the solar radiation that plants use for photosynthesis; technically referred to as Photosynthetically Available Radiation (PAR; Kirk 1994). Sunlight is absorbed by water, even when it contains little suspended sediment (Kirk 1994). Light intensity decreases with depth in an exponential manner, the rate of which is dependant upon the turbidity of

146

the water and the wave-state of the surface (Kirk 1994). Corals at deeper sites are expected to receive less light for growth than corals at shallower sites. Sites at a particular depth in water with higher turbidity should likewise receive less light than sites at the same depth in clearer water.

The Effect of Aspect The coral reefs in Bermuda represent the most northerly coral reefs in the Atlantic. As such, the angle of the sun in the southern sky is more acute here than on reefs located closer to the equator (Figure 3.06). For this reason, the southern side of a patch reef in Bermuda should be exposed to substantially more sunlight over the course of the year relative to the northern side of the same patch reef. Below I describe a study (Project 2) to determine the differences in light flux to a sensor facing a southerly vs a northerly direction. In Chapter 4 I compare the differences in the coral assemblages located on the north to those on the south sides of the same patch reef at the same depth.

147

Figure 3.06. A graph of the hourly positions and paths the sun appears to take as it crosses the sky in Bermuda over the course of a day during the summer and winter solstices, and either equinox. The numerals around the circumference of the circle represent bearing, with 90° representing East. The numerals within the circle represent the angle above the horizon, with 90 representing the zenith point, directly overhead. The sun is at it’s apparent daily apogee in Bermuda at 12.13 pm local standard time. Graph generated using the shareware tool “Sol Path” (© C. Gronbeck 2002, 2006).

148

Wave energy and currents The island of Bermuda possesses a hilly topography with an average height of 30 m above sea level (Morris et al. 1977), which acts to block the effects of the wind on marine habitats located in a leeward direction. As the island is located along the south-east edge of the lagoon, the island primarily blocks winds that originate from the east through to a south-westerly direction from reaching lagoonal waters. The prevailing winds in Bermuda are from a south-westerly direction throughout the summer (Appendix C, below), which results in calmer waters being found near the northern shore and higher waves occurring offshore. In the winter, the prevailing winds originate from the northwesterly direction. As there are no islands to block the wind from the north-west, high seas produced by winter storms affect all lagoonal reefs to a relatively equal degree (Morris et al. 1977 ), and even the northern shore is subjected to large (>4-m) waves (Thomas 1985). Additionally, wind-driven current flow declines with depth, at an exponential rate (Ekman 1905). Bermuda experiences a semi-diurnal tidal cycle with a range of 0.75 m on average (Morris et al. 1977). As a result of the changing tides, strong currents occur in some areas on the Bermuda platform, particularly at the entrances to enclosed bodies of water such as Harrington Sound and the Great Sound (Figure 3.02; Morris et al. 1977). Strong tidal currents can also be observed to flow over the lagoonal rim at all locations, although they are strongest at the northeast and southwest ends of the lagoon, presumably due to the position of the Bermuda islands. However, since the study area is located within the central area of the Bermuda lagoon,, far from enclosed bodies of water and the NE and SW parts of the lagoon, tidal currents have little strength.

149

Project 1: Mapping of lagoonal reefs Introduction and Methodology When the project was first initiated there were no accurate maps available that delineated the locations of the patch reefs found scattered within the lagoon. For this reason, many of the sites I surveyed were selected based on the surprisingly inaccurate maps of reef location available from the Hydrographic Office of the United Kingdom, as well as from triangulation of terrestrial landmarks and extensive time spent out in the lagoon in boats learning the “lay of the land”. In order to aid in the scientific study of Bermuda’s lagoonal reefs henceforth, in 2003– 2004 I produced the first accurate, geo-referenced digital map of the entire suite of reefs visible within the study area, as part of a larger mapping effort I did as a member of the Bermuda Zoological Society that included all lagoonal patch and pinnacle reefs. This map was produced by referencing a mosaic set of georeferenced aerial photographs of the islands and surrounding submerged platform that the Bermuda Zoological Society commissioned in 1997 (Figure 3.01). The aerial photographs of the Bermuda Islands and surrounding reef platform were produced using a Zeiss Jena LMK photogrammetric survey camera with forward motion compensation, which was mounted onto a small aircraft. In 2003 a composite raster bitmap for the entire Bermuda platform was produced from the slides, with a final resolution of 50 cm per pixel and a geo-referenced error of less than 2 m (Bermuda Zoological Society 1997). A map of probable coral reef habitat was then created for the entire Bermuda platform from the digital mosaic with the GIS package ESRI ArcMap 9.1

150

(Figure 3.02). To create this map I manually digitized the boundaries of coral reef areas as continuous closed polygons at a scale of 1:2500, using color (light to dark reddishbrown) and the presence of sand halos around reefs as visual indicators of boundary edge. Spatial referencing of the digital photographs was accurate to 2 m and, combined with a pixel size of 50 cm, a spatial accuracy of about ±2.5 m was possible for visually mapped reef boundaries. Over 34,000 separate reefs were mapped across the extent of the lagoon. Once the boundaries of each reef within the lagoon was mapped to GIS, I delineated the outer edges of the study area, as well as the margins between each of the six zones within the electronic map. I then used the Hawth’s Analysis Tools extension for ESRI’s ArcGIS (Beyer 2004) to generate the data which I compiled into Table 3.01, which details the area enclosed by each zone, the number of reefs in each zone, the density of reefs per km2, the total reef area per zone, the percentage of total zone area covered by reefs, and the mean size of each reef.

151

Figure 3.07. Zonal boundaries, locations of the north and south shipping channels, and location of patch reefs distributed across the study area encompassing the North Lagoon. The numbers allocated to each zone are written on the right side of the map.

152

Results and Discussion The characteristics of the patch reefs within each of the six zones surveyed are listed in Table 3.01. Zone 1 lies closest to shore and is bounded on the southern side by the northern shore of the island. The northern boundary is delineated by the South Shipping Channel which cuts across the study area. Zone 2 is located on the northern side of the South Shipping Channel and extends 1-2 km to seaward. Zones 1 and 2 had the fewest reefs per km2, but the mean reef size was larger than most other zones. Many of the nearshore reefs in Zone 1 appear to be fringing reefs, which are large, linear structures that are positioned with the long axis running parallel to shore. These reefs probably formed on a basement of drowned shoreline and only formed in the past 4,000-6,000 years (Garrett et al. 1971; Garrett and Scoffin 1977). The reefs in Zones 2 to 4 are mostly contained within the clusters of reefs known as Baileys Flat and Brackish Pond Flat. These reef systems appear to have formed on fossilized sand dunes that formed during glacial periods when the shallow platform was subaerially exposed. Between these two large reef clusters is a deeper area of approximately 12-m depth across which are scattered many smaller patch and pinnacle reefs, which may contribute to the lower mean area of the reefs in the middle zones of the platform. The reefs in zones 5 and 6 are large faro reef complexes, formed on what are most probably wash-over sediment deposits that formed 8000 to 6000 years ago when the platform first flooded (Garrett et al, 1971; Garrett and Scoffin 1977). These reefs form large clusters with large reefs surrounding a shallow inner sand-covered area speckled with small patch reefs.

153

Table 3.01. Characteristics of each zone of the survey area across the Bermuda Lagoon, and the patch reefs contained therein.

154

Project 2: Assessing the three-dimensional light field over a range of depths Introduction and Methodology Although light in oceanic waters can be accurately quantified as a one-dimensional gradient of downwelling and upwelling flux, light in shallow coastal environments has a three-dimensional character (reviewed in Kirk 1994; Ackleson 2003). Bermuda’s high latitude further enhances the multidimensional character of the light field, in that the angle of the sun in the sky is significantly sharper through much of the year than it is at all other coral reef regions, which al lie further to the south. To quantify the three-dimensional character of light in Bermudian waters, I assessed the differences in light flux measured by a hemispherical sensor, representing a moundshaped coral colony, when it was positioned perpendicular to the directions North, East, South, West, Up and Down (N, E, S, W, U, D). To do this I modified a standard scalar LiCor light sensor, which is spherical and designed to measure light impinging from all directions, by covering the bottom half of the sphere so that light could only be received over the top half of the sphere (Figure 3.08). Light was blocked from the bottom half of the sensor with the use of opaque adhesive tape, and the entire sensor was mounted at the midpoint of an circular and opaque black plastic collar with a 25-cm diameter. The collar was added to further prevent the reception of light by the sensor from sources located below the exposed hemispherical sensor surface. I measured light availability from the underwater light field emanating as downwelling light, upwelling light, and light impinging from easterly, southerly, westerly and northerly aspects. Average measurements from these six directions were recorded by the data logger at 5-second intervals for 5 minutes each at 3-m depth intervals down to

155

15-m depth. Data collection was done in clear oceanic water (Secchi disk depth >18 m) at location 32.382°N, 64.648°W on the forereef east of Bermuda, on July 24th, 2002. A second light sensor was placed at 9-m depth to control for the effects of clouds. Concurrent measurements taken in air and at 9-m depth prior to the experiment were used to convert recorded light flux to a percentage of the light hitting the surface of the ocean. Both sensors were connected by transmission cables to a electronic data logger located on the boat at the surface. The boat was not directly overhead of the area of data collection, so as to avoid detection of the boat shadow. The mean light flux and standard error of each collection period from each depth and aspect were subsequently graphed for visual comparison. The data did not conform to the assumption of equal variances, and transformations of the data were unable to resolve the problem. The results of ANOVA on all forms of the transformed data produced the same significance values as did individual Mann-Whitney U tests (a nonparametric test of differences), so the results from the ANOVA using non-transformed data are presented.

Results and Discussion At all depths the amount of light impacting the sensor when it faced a southerly direction was substantially higher than when it faced a northerly direction (Figure 3.09). Downwelling light was substantially higher than the amount of upwelling light (as expected), and downwelling light was also significantly higher than light originating from either a southerly direction or a northerly direction. Additionally, more downwelling light was available in locations closer to the surface than at greater depths. The values for the sensor placed in an easterly and westerly aspect were statistically indistinguishable from

156

each other and at levels between those obtained from the northerly and southerly aspects, which is as would be expected (T. Murdoch unpublished data). For simplicity, the data from these two “midpoint” directions have been omitted from further analysis in this section. Two-way ANOVA of depth and aspect for all 5 depths and four directions (N, S, U, D) found that the interaction between the depth and the direction that the sensor faced was highly significant at p < 0.001 (Table 2.02). Tukey HSD post hoc comparisons of the light data determined that all measurements were significantly different for each depth and aspect except between all measurements taken at 12-m depth and at 15-m depth, and between the measurements taken from the northerly aspect and of upwelling light at all depths. These results illustrate that the location of Bermuda at a high latitude results in more light being available to an organism living on the south-facing side of an object, such as a coral reef, than on the north side, due to a shading effect created by the bulk of the reef itself. The survey was done in August, when the angle of the sun is more obtuse than it is for eight months of the year. Thus, for most of the year the relative difference in light availability between the north and south sides of a reef is greater than that observed. However, since less light reaches the surface of the earth as the sun angle sharpens, the absolute difference between the northern aspect and southern aspect would be less in winter months.

157

Figure 3.08. A diagram illustrating the modified Li-Cor scalar PAR sensor. Opaque adhesive tape, an opaque plastic jar, and a black plastic collar constrained the light quantified by the sensor to that impacting the upper, exposed hemisphere.

158

Figure 3.09. Mean proportion of surface light (± standard error) originating from four directions at five depths, as measured by the hemispherical sensor.

159

Table 3.02. Two-way analysis of variance of the effects of depth and aspect on the proportion of surface light reaching a hemispherical sensor. Source Depth Aspect Depth * Aspect Error

SS 5382523.974 24710913.276

df 4 3

MS 1345630.993 8236971.092

F-ratio 178.307 1091.466

p .000 .000

5931307.572

12

494275.631

65.496

.000

2867746.630

380

7546.702

160

Project 3: Cross-platform differences in downwelling light availability Introduction and Methodology If turbidity is higher nearshore relative to offshore, as prior research (CARICOMP 1999; Jones 2007) indicate, then the amount of downwelling light impacting a point at a particular depth should be lower nearshore than offshore. In order to assess crossplatform differences in the availability of downwelling light, light flux was assessed at 8m depth at five equally-spaced locations that were at different distances from shore (Figure 3.10), over a week-long interval in June 2004. Downwelling light intensity was measured using a collection of miniaturized, submersible Onset Hobo® light sensors with self-contained data loggers, following techniques described in Kirk (1994). Each sensor recorded light flux at 5-min intervals over the 7-day period. The maximum light flux readings taken over the 2 hr in the middle of each daylight period were used to calculate differences between locations in light availability. Only data from the first five days were used, as most sensors became fouled with fine sediment or filamentous algae after that period of time. Transformed data did not conform to the assumptions of homogeneity of variances; the results from ANOVA of the nontransformed data are presented but should be interpreted with caution.

161

Figure 3.10. Map of the five reef sites, indicated by the light-bulb symbol. At each site luminance readings were taken concurrently, from 26 July 26 to 1 August 2004. All sensors were placed at exactly 8-m depth, with the sensor surface facing straight upward. The boundaries of the study area, the six zones, and the north and south shipping channels are also illustrated

162

Results and Discussion Figure 3.11 illustrates the continuous changes in light level over 5-min intervals over the five days across the five sites. Light levels increase over the course of a day until 12:13 pm local standard time, and then decline again as the surface of the earth rotates away from the sun. Sharp spikes in light above the daily maximum are characteristic of peaks in light produced by the passage of waves. A surface wave can act as a lens, and may either focus or diffuse light reaching a point below it (termed “wave focusing”; Kirk 1994; Schubert et al. 2001). This process of wave focusing is the source for the meandrous patterns of brighter light observed on the sandy sea floor in shallow water, or on the bottom of a swimming pool. The magnitude of the spike during wave focusing events is dependent on the shape of the wave and the depth of the sensor, with the highest spikes occurring when lens shape and sensor depth match and a large amount of the sun’s light is focused on the spot. Sudden large drops in light intensity that one can see in the daily light data are probably due to the occlusion of the sun by clouds (Kirk 1994). Drops in light intensity that match across all sensors, such as the event recorded near noon on the third day surveyed, indicate the passage of very large clouds that encompassed the entire reef platform. Another source of reduced light across large areas is the occurrence of sustained wind events with speeds greater than 28 km/hr (15 kts), which cause the formation of frothing waves or “white caps” that reflect light away from the seabed (Kirk 1994). Light levels were highest at the site located in the rim reef zone, and declined at sites progressively nearer to shore (Figs. 3.11, 3.12). ANOVA determined that the overall

163

pattern was highly significant (Table 3.03). Tukey’s HSD post hoc analysis (Table 3.04) determined that only sites C and D were not significantly different in light intensity. These results indicate that, for a given time period at a given depth, corals nearshore are exposed to less light than those offshore. Since light is a critical resource for hermatypic corals which host photosynthetic zooxanthellae (Rogers 1979; Achituv and Dubinsky 1990), and since different coral species are adapted to cope with a range of characteristic annual light budgets (Huston 1985a; Graus and Macintyre 1989), it is expected that different corals will be found at the same depth but at different distances from shore across the Bermuda platform.

164

Figure 3.11. Light intensity readings taken over five days from sensors positioned at 8-m depth at 5 reef sites located at different distances from shore across the area of study. Instantaneous light intensity readings were taken at 5-second intervals over five days.

165

Figure 3.12. Average light intensity (Lumens ±SE) measured from 11 am to 1 pm local standard time over the first of the five days of deployment, by light sensors located at 8-m depth at five reef sites positioned at increasing distances from the North Shore of Bermuda. The approximate distance from shore of the boundaries between zones (which varied along the extent of each zone) are also indicated.

166

Table 3.03. ANOVA table of the differences in light intensity across the five locations from the data illustrated in Figure 2.13. Source Distance Error

SS df 193031.858 4 25626.866 115

MS 48257.964 222.842

F ratio 216.557

p 0.000

Table 3.04. Results of a Tukey’s post hoc analysis of the significance in the differences in the amount of luminance at 8-m depth over the hours of 11 am to 1 pm between the five locations across the reef platform. Only sites C and D did not differ significantly in the amount of light measured at 8-m depth.

Tukey HSD Distance A B C D E

A

B

C

D

0.000 0.000 0.000 0.000

0.009 0.000 0.000

0.596 0.000

0.000

167

E

DISCUSSION A review of the literature regarding Bermuda’s physical oceanographic conditions, as well as novel research, indicates that the four critical environmental factors of temperature, light, suspended particulate matter and current flow all covary with distance from shore and with depth. As described in Chapter 1, these four factors can each affect corals as stressors, disturbance agents and as factors promoting growth, depending on the levels or intensities at which they occur (Figure 3.13). The amount of variability in temperature is higher nearshore compared with offshore, with colder winter temperatures and hotter summer temperatures experienced by sites near land (Morris et al. 1977; du Putron 2003). Offshore sites, which are bathed in the waters of the surrounding Sargasso Sea, experience less change in temperature over the year. Temperatures also vary across depths in the summer, with warmer waters occurring at the sea surface across the lagoon (Morris et al. 1977). Low temperatures inhibit physiological functions, and as such act to inhibit growth and tissue repair. High temperatures, alternatively, act as a disturbance agent, by inducing coral bleaching (Cook et al. 1990; Brown 2004), inhibiting reproduction (Fraser and Currie 1996), promoting the growth of coral diseases (Harvell et al. 1999), and the growth of competing organisms such as macroalgae (Johannes et al. 1983). Accordingly, nearshore reefs are predicted to have less coral cover and species relative to offshore reefs. Additionally, competitively-dominant corals may be expected to dominate in areas of lower temperature range found offshore, with disturbance-tolerant or stress-tolerant corals dominating the nearshore areas with a high range of temperatures. Light levels in Bermuda are highest in the shallowest waters offshore, and decline

168

both with depth and with proximity to shore. Wave energy follows the same pattern. Both light and wave energy act as damaging agents at high levels, promote growth at moderate levels and inhibit productivity at very low levels. For that reason it may be expected that ruderal corals dominate shallow sites, particularly on reefs offshore, while competitivedominate coral species dominate at some mid-point across the lagoon and at moderate depths. Deep sites, particularly nearshore, are predicted to be dominated by stress-tolerant coral species adapted to low levels of light and low rates of dissolved nutrient flux. Suspended particulate matter (SPM) is in highest concentration nearshore, and decreases in concentration as the distance from shore increases (Mills et al. 2004; G. Toro-Farmer unpublished data). SPM can act as both a disturbance agent and as a source of nutrients. Nearshore sites are predicted to possess more corals that are capable of removing or surviving the smothering effects of SPM. Additionally, heterotrophic corals that are able to utilize SPM instead of light as a source of carbohydrate nutrition may be favored in nearshore habitats relative to those offshore. In Chapter 4 I describe how I examined the manner that species and functional groups of coral varied across sites that differed in distance from shore and depth, which serve as simple proxy direct gradients for the co-varying physical gradients of temperature, light, SPM and wave energy that were described in this chapter. Coral assemblages were surveyed across sites located at a range of depths and distances from shore on patch and pinnacle reefs within the North Lagoon of Bermuda, and the results compared to the patterns predicted by the modified Adaptive Strategies Theory that I presented in Chapter 1.

169

Figure 3.13. Representation of how varying levels of the four most important environmental factors act to promote growth, or act as a stressor or disturbance agent to corals.

170

171

CHAPTER 4: THE DISTRIBUTION OF CORAL SPECIES AND FUNCTIONAL GROUPS OVER PHYSICAL GRADIENTS ACROSS THE NORTH LAGOON OF BERMUDA

INTRODUCTION

According to the Adaptive Strategies Theory (AST; Grime 1979), one can predict the characteristics of the functional group of organisms that will dominate a particular habitat by determining the levels of stress and disturbance that are found at that location (Figure 4.01). On coral reefs, the four environmental factors that predominantly affect the amount of stress or disturbance which corals experience are: water temperature, solar radiation (i.e. sunlight), suspended particulate matter (SPM) and water flow. In Chapter 3 I described how these four physical factors vary as gradients across the North Lagoon of Bermuda. In this chapter I examine whether the modified AST that I described in Chapter 1 predicts how species and functional groups of coral are distributed across sites on reefs in Bermuda that are located over these environmental gradients of stress and disturbance. The sites I surveyed in Bermuda can be thought of as representing the range of environmental conditions enclosed within the nested square illustrated in the AST (CSR) diagram below (Figure 4.01).

172

The four critical environmental factors of temperature, light, suspended particulate matter and current flow all vary with distance from shore and with depth (Chapter 3). As described in Chapter 1, these four factors can each act on corals as either stressors, disturbance agents and in the promotion of growth, depending on the levels or intensities at which they occur (Figure 1.06). Sea water temperature exhibits a wider range within the surveyed area at nearshore sites compared with locations offshore, with colder winter temperatures and hotter summer temperatures experienced by sites near land (Morris et al. 1977; de Putron 2003). Since corals suffer physiological disturbance when they experience temperatures too far outside their average range, nearshore reefs are predicted to have less coral cover and species relative to offshore reefs. due to high disturbance levels (strong annual seasonal temp fluctuations) restricting the growth of stress-tolerant corals and limiting the number of species capable of surviving on these nearshore reefs. Additionally, competitivelydominant corals may be expected to dominate in the areas of lower temperature range found offshore. Light levels in Bermuda are highest in the shallowest waters offshore, and decline both with depth and with proximity to shore, due to the re-suspension of sediments. Wave energy follows the same distribution pattern. Light and wave energy both damage corals when at high levels . At moderate levels light and water flow both promote coral growth. At low levels of both light and water flow corals experience limited productivity . Accordingly, ruderal corals are predicted to dominate shallow sites, particularly on reefs offshore, while competitive-dominate coral species are predicted to dominate at some mid-point across the lagoon and at moderate depths. Deep sites, particularly offshore, are

173

predicted to be dominated by stress-tolerant coral species adapted to low levels of light and low rates of dissolved nutrient flux due to reduced rates of water flow. Nearshore reefs may see high nutrient fluxes due to the presence of nitrogen-rich ground water. Suspended particulate matter (SPM) is in highest concentration nearshore, and decreases in concentration as the distance from shore increases (von Bodungen et al 1982; Mills et al. 2004; G. Toro-Farmer unpublished data). SPM can act as both a disturbance agent and as a source of nutrients Nearshore sites are predicted to possess more corals that are capable of removing or surviving the smothering effects of SPM. Additionally, heterotrophic corals that are able to utilize SPM instead of light as a source of carbohydrate nutrition may be favored in nearshore habitats relative to those offshore.

OBJECTIVES

With the use of direct gradient analysis I tested the following hypotheses regarding how species and functional groups of coral varies across the sites surveyed in the Bermuda lagoon:

Similarities in the distribution of species and functional groups Irrespective of the manner in which stress or disturbance agents are distributed across sites, species within functional groups are hypothesized to have similar functional responses to the environment. If this is so, then species belonging to the same functional group should differ little in their distributions across sites, due to the shared responses to environmental and biological conditions (Steneck and Dethier 1994; Gitay and Noble

174

1997; Hooper et al. 2002). Conversely, the distribution patterns of species belonging to different functional groups should be substantially different from each other, due to disparate environmental tolerances. Chapter 1 details how these differences in site fidelity should appear when distributional data is graphed in a variety of ways. Despite my best attempts to do otherwise, it may be that the traits I selected to use in the categorization of species into functional groups are those indicative of functional mechanisms by which species acquire resources, and not those which are used to cope with differing environmental disturbances. If this is so, then species within functional groups may overlap greatly in their nutritional needs and therefore compete strongly amongst other members for resources Species that belong to the same resource-based functional group (also known as “guild”) should tend to occur in different environmentally-defined habitats (reviewed in Fox 1999). Also species that are members of different guilds should be able to coexist within a location to a greater degree than species from the same guild.

175

Figure 4.01. A modified version of Grime’s (1977) and Steneck and Dethier’s (1994) generalized two-dimensional AST model of FG dominance within habitat types, incorporating the concession that biota can only survive in habitats within which the rate or amount of resource acquisition (resource abundance) is greater than the rate or amount of resource loss (or disturbance). The boundary between the white and grey areas is the zero net growth intercept (ZNGI) of the assemblage as a whole. The nested square within the AST diagram of state space represents the hypothetical range within the AST model that was represented by the sites surveyed within the north lagoon of the Bermuda Reef Platform. Letter designations are: C – Competitive Dominant; R – Ruderal; S – Stress tolerant.

176

Similarity among sites in species assemblages Sites were compared according to the degree to which they shared coral species. Sites that differed in aspect, depth and distance from shore were evaluated so that the species composition of each site could be interpreted accordingly. Sites that shared environmental characteristics, regardless of zone or depth, were expected to cluster together in multidimensional species-abundance space. Sites at a similar depth and distance from shore were expected to cluster together in multidimensional state space (i.e. have similar assemblages of corals on them) because the shared environmental conditions of each site should filter coral species membership in the same way.

Percent cover and abundance per functional group The percent cover for corals overall, and for the group of species that are members of the most competitive functional group, are both predicted to peak on reefs where the levels of disturbance and stress are lowest. In the north lagoon light and wave energy are highest at shallow, offshore sites and lowest at nearshore, deep sites. Conversely, SPM and temperature are most damaging to corals at nearshore locations regardless of depth, and decline with distance from shore. Accordingly, moderate levels of all four variables are expected to occur at mid-depth and roughly in the centre of the lagoon. Since moderate levels of the four variables are the least limiting to corals, the average coral cover of all species, as well as of the competitive species should peak at sites located in the middle of the lagoon as well. Members of the Branched, Oviparous (BO) functional group (i.e. the acroporids) are not found in Bermuda. With the BO group absent, either the Competitive-Ruderal or Competitve - Stress-tolerant functional groups may plausibly

177

occupy its environmental niche, and thus dominate habitats characterized by low levels of stress and disturbance in Bermuda. Branched Viviparous (BV) species represent the Competitive-Ruderal group. Species of the Massive Oviparous (MO) functional group belong to the Competitive-Stress-tolerant group. Massive, viviparous (MV) corals, which I classified as Ruderal according to the modified AST, are predicted to dominate at offshore reef sites in shallow water where disturbance levels are predicted to be the highest. Stress-tolerant corals of the Foliose and Plating (FP) functional group are predicted to be intolerant of disturbance at all but the lowest levels. Since temperature variability and SPM, which are both forms of disturbance, are higher nearshore compared with offshore over all depths, stress-tolerant corals should dominate only deep sites located in the zones furthest from shore.

Species distributions across sites In addition to providing information regarding the percent cover and relative abundance of each functional group across all sites, as I did in Chapter 2, I also determined the distribution patterns of each species. This species-specific data allows one to see the degree to which species within functional groups share distributional patterns, if at all. I also present this data for a pragmatic reason. This project represents the most intensive survey of reef sites across Bermuda’s north lagoon to date, and the only survey of how coral assemblages change with depth and distance from the shore across patch reefs across the lagoon. For this reason, species distribution data are provided in order to aid in management and conservation of the corals, as well to guide future scientific research.

178

Species richness of all corals According to the predictions of the productivity-diversity hypothesis, which lead to CSR theory (Grime 1973; 1979), and the intermediate disturbance hypothesis (Connell 1978; Aronson and Precht 1994; Huston 1996), the habitat types with intermediate levels of stress and disturbance should display greater species richness than either the habitats with low levels of stress or disturbance and the habitats with high levels of stress or disturbance. Sites located in the middle of the lagoon and at mid-depth are thus predicted to possess the highest number of species, with lower numbers of species found at the extremes of distance and depth.

Functional group richness Grime (1977) predicts that the richness of functional groups will be low across sites, as he predicts each functional group will be replaced by another from site to site due to the combined effects of environmental filtering and competitive exclusion between functional groups. Alternatively, if competition does not operate between groups, then all functional groups are predicted to occur at mid-depths on reefs centrally located within the lagoon, as these reefs are characterized by moderate levels of the four environmental factors. Shallow sites offshore and all sites inshore are predicted to not possess the stresstolerant functional groups, and deeper sites should not possess the ruderal nor competitive-ruderal species, according to the AST.

179

METHODOLOGY

Data collection The tops of three replicate patch reefs (designated E, C and M), and the sides of two of the three patch reefs (E and C), separated by at least 1-km, were sampled within each of the 6 zones (Figs. 4.02, 4.03; Table 4.01). Zones differed in distance from the North Shore of Bermuda and were within the lagoonal area bounded by the rim reef to the north. At each reef, transects were videotaped following the procedures described in Aronson et al. (1994), Aronson and Swanson (1997), for fore-reef habitats, but with modifications that were needed in order to cope with the different geomorphology of patch reefs. Since variation in the size of reefs may have affected the composition of the coral assemblage in unintended ways (Keough 1984), all of the reefs surveyed were of similar area and overall shape. Each patch reef was oval in shape, roughly 60-m long, 40-m wide, and with the long-axis lying in an east-west direction. The top of each patch reef was assessed using eight 10-m long transects, placed haphazardly and so as to encompass as much of the top of the patch reef as possible Only eight transects were used as no more could be surveyed without overlap. . Additionally, four 10-m long transects were haphazardly placed horizontally along consecutive ~2- to 3-m depth contours on the southern and northern sides of the reefs.. Only four 10-m transects were filmed on each flank; this representing the maximum number of transects that could fit along the ~ 50-m long flank of each patch reef at each depth, while still allowing for the haphazard placement of the transects. Since the four transects encompassed most of the surface of the reef at each depth it seemed probable that the sampled population

180

represented most or all of the total population of corals found at each site, which is to say that most or nearly all of the statistical universe was recorded. All transects were videotaped along a 0.4-m wide x 10-m long swath of the reef with a high-resolution digital video camera enclosed in an underwater housing. An aluminum bar projecting forward from the camera housing was used to maintain a 40-cm distance between the camera lens and the reef surface. A depth gauge mounted on the end of the bar displayed the depth of the reef substrate on each video frame, providing a measure of the rugosity of the reef surface when filming the tops of reefs. The depth gauge also (1) aided the videographer in maintaining a set depth while filming transects on the steeply sloped sites of pinnacle reefs, (2) provided scale in the videotaped images, and (3) provided a record of the depths surveyed on each reef.

181

Figure 4.02. General design of the study, in which survey sites (circles) were surveyed at a range of depths on the north, south and top sides of replicate patch reefs within each of six zones located at increasing distances from shore. Not shown are the replicate reefs within each zone.

182

Figure 4.03. A map of the lagoonal reefs located within and around the research area and the 18 patch reef sites surveyed in the videographic analyses. Horizontal lines indicate the six zone boundaries. Grey lines represent the location of the north and south shipping channels. A portion of the island of Bermuda is represented by the area in darker gray at the bottom of the image.

183

Table 4.01. Details of the 18 surveyed reefs surveyed across the North Lagoon, also mapped on Figure 4.03. Replicates labeled M were only videographically sampled on the top of the reef, whereas replicates E and C were surveyed on the tops as well as across a range of depths on the north and south sides of each reef. Reef Top Zone

Replicate

Name

Site

Lat

Long

Depth (m)

1

E

Baileys Bay Patch

E1

32.3520

-64.7250

7

1

M

Shelly Bay Patch

M1

32.3290

-64.7440

7

1

C

Robbie’s Reef

C1

32.3170

-64.7520

7

2

E

Inner Baileys Flat

E2

32.3530

-64.7480

5

2

M

Shelly Bay Shoals

M2

32.3360

-64.7570

5

2

C

Finger Coral Patch

C2

32.3356

-64.7765

7

3

E

Martello Patch

E3

32.3720

-64.7430

6

3

M

Judie's Awakening

M3

32.3700

-64.7620

6

3

C

Gaeroid’s Reef

C3

32.3550

-64.7820

7

4

E

Angel Reef

E4

32.3808

-64.7660

4

4

M

Jeannette's Reef

M4

32.3750

-64.7830

4

4

C

Sascha’s Reef

C4

32.3643

-64.8062

4

5

E

8A Hole Reef

E5

32.4066

-64.7830

4

5

M

S. Crescent 3 Patch

M5

32.3870

-64.7990

3

5

C

Pawlik’s Reef

C5

32.3940

-64.7890

2

6

E

Pretty Bumpy Reef

E6

32.4180

-64.8193

2

6

M

Lisa's Reef

M6

32.3961

-64.8605

2

6

C

Claire's Reef

C6

32.4094

-64.8410

3

Data analysis

184

Different hypotheses were tested using different types of data derived from the videotaped transects. Relative Frequency of occurrence (RF) data for corals was used for the analysis of the similarity in species distributions, the similarity of sites in species composition, and the frequency of occurrence of each species across all reefs, as well as species richness and functional group richness data. RF data was used for these analyses instead of point-count data as it included rare or small species that may be missed in the assessment of randomly placed points on transects. The RF data were collected in the following manner. 20 non-overlapping video frames were digitally captured from each of the 10-m long transects. Following image capture, the presence or absence of each species in each of the 20 frames was recorded. The proportion of frames over which each species occurred in each transect was then calculated by dividing the number of times each species was present in each of the 20 frames. Point-count data was assessed from each video transect following a computerautomated version of the procedures described in Aronson et al. (1994), Aronson and Swanson (1997) and Murdoch and Aronson (1999) that were developed by the author for a different project (Appendix 4B). The video transects from the tops of three replicate reefs, and the tops and southern flanks over a range of depths for two replicate reefs for each of the six zones were assessed for the percent cover of individual coral species by point-counts Twenty-five random points positioned on each adjacent digital image captured from the videographic tape. The alternate (odd-numbered) frames were captured to provide other views of the same substrate and as an aid for visual analysis of point count locations but were otherwise not analyzed. . The results of each analyzed frame

185

were summed for the entire transect, and the average of the four or eight transects was used to calculate the average percent cover for each of the sites on a reef.

The following variables were calculated for each reef site from the videographic data:

Species coexistence among habitats The frequency of occurrence (RF) data were used to calculate a similarity matrix for the suite of species, based on the proportion data from each transect and separately based on the proportion data averaged at each site. Transect-level and site-level results were alike, so the less complicated site-level analysis is described. A Bray-Curtis similarity metric was calculated for each species pair from square-root transformed proportional data. Since frequency of occurrence data exhibits a smaller range of values than biomass data, it need only be square root transformed, instead of fourth-root transformed, when used to generate similarity matrices (Clarke and Gorley 2006). Using the Bray-Curtis metric, species that exactly shared the same distribution pattern were assigned the highest percent similarity (i.e. 100%), and species with distribution patterns that did not overlap at all were assigned very low similarity values. The benefit of looking at the coexistence of species within and among functional groups in this way is that knowledge of the actual environmental and biological conditions of each site was not needed, as long as sites were sampled across a broad range of habitats.

186

Similarity among sites in species assemblages Sites were also compared according to the degree to which they shared coral species using the same frequency of occurrence data as described above. Sites that differed in aspect, depth and distance from shore were compared so that the species composition of each site could be interpreted accordingly. Sites that shared zone and depth were expected to cluster together in multidimensional species-abundance space, because sites with similar depth and distance from shore shared environmental conditions and therefore should affect coral species membership in a similar manner.

Species distributions across sites In order to see how member species of each functional group were distributed across sites in the four dimensions surveyed (i.e. RF x depth x aspect x replicate reef), I graphed the frequency of occurrence data for each species. These data were averaged for each of the four or eight transects for each site and the results graphically presented in three ways. An MDS diagram was generated for all sites, and the average RF of each species and functional group plotted as a circle, with differing sized representing differences in occurrence. Additionally average (±SE) RF was plotted as points on a line graph for each replicated site, aspect, depth and reef. The same data were also plotted onto a bubble diagram for each depth and reef (but not aspect). Bubble diagrams present the relative abundance for each species across sites and depths in a two-dimensional manner that is analogous to the actual distribution of sites across real space. These three disparate graphical techniques were concurrently used for the identical data so that the reader could

187

better comprehend the 4D manner in which each species and functional group was distributed across sites. Percent cover data for each species at each sites was also used to determine the distributional patterns of each species across the tops of three replicate reefs and tops plus south sides of two replicate reefs in each zone. The method by which these analyses were done is described below.

Standard measures Percent cover, species richness and functional group richness measures of the coral assemblage were also collected across sites from video transects. This data was collected from video transects filmed on the tops of three replicate reefs and on the tops and south sides of two replicate reefs located in each of the six zones. The information from the tops and south sides was collected for two reasons: (1) to test the hypotheses of the modified AST using the kind of information usually collected by coral-reef scientists in the habitats where research is generally focused (i.e. the tops of reefs and not the sides); and (2) to accurately determine the manner in which coral parameters vary across the North Lagoon. The data from the north sides were ignored so as to reduce the complexity of the analysis by ignoring the effect of aspect.

188

Statistical analyses Codes for species and functional groups The acronyms used in the graphs to follow for each functional group are: BV: Branched Viviparous; FP: Foliose and Plating viviparous MV: Massive Viviparous; MO: Massive Oviparous;

Functional group membership used in the graphs to follow, and the letter designations for each species are as follows: Functional Group

Species

Abbreviation

BV

Madracis decactis

MDEC

BV

Madracis mirabilis

MMIR

BV

Porites porites

PPOR

FP

Agaricia fragilis

AFRAG

MO

Diploria labyrinthiformis

DLAB

MO

Diploria strigosa

DSTRIG

MO

Montastraea cavernosa

MCAV

MO

Montastraea faveolata

MFAV

MO

Montastraea franksi

MFRANK

MO

Stephanocoenia intersepta

STEPH

MV

Favia fragum

FAV

MV

Porites astreoides

PAST

MV

Siderastrea radians

SRAD

A Bray-Curtis similarity matrix was generated for all observed coral species, based on their distributions across all reef sites. The results of the matrix was graphically

189

displayed, as both a dendrogram and a multidimensional scaling (MDS) diagram (Figs. 2.19-2.23) An analysis of similarity was calculated in order to determine whether functional groups of species formed significantly distinct clusters. A two-way analysis of variance was calculated for each of the standard univariate measures assessed.

190

RESULTS

Depths per reef Figure 4.04 illustrates the number of sites per reef and the depths at which each site was surveyed on each patch reef. The number of sites per reef was limited by the size of the flanks (i.e. sides) of each reef, and reefs in different zones had flanks of different heights. Reefs in Zones 1 and 2 were small and deep and therefore only one depth could be surveyed per side. Reefs in Zones 3 and 4 were in shallower water, but reached closer to the surface and as a result two could be surveyed per side. Reefs in the outer two zones were in deeper water, but reached the shallowest depths as well. As a result, three depths could be surveyed on the reefs in zone 5 and five depths surveyed down each flank on reefs in Zone 6. The reefs in most zones were patch reefs with sloping sides. Reefs in Zones 1 and 2 did not contain deeper central basins of sediment. Reefs in Zones 3 –5 were better developed and did contain a deeper central “mini-lagoon” containing sediment and limited coral development, These central areas were avoided in the transects. The reefs in Zone 6 were well-developed pinnacle reefs, with shallow, flat tops without central mini-lagoons, and had very steep or overhanging sides reaching to 12-m depth. The limited height that characterizes the reefs in Zones 1 and 2 may be indicative of a highly disturbed and stressed environment to corals. These two zones are separated by the southern shipping channel, in which light levels are reduced and suspended sediment loads are high due to the passage of container and passenger ships throughout the year.

191

Figure 4.04 should be referred to when examining the graphs to follow, as the manner in which sites graphically arranged on each reef are the same manner throughout. In each matching graph, the site located at the top of each reef is within the central white band for each zone, while south-facing sites are illustrated within the light gray band in each zone, and the sites with a northern aspect are illustrated within the darker gray band in each zone.

192

Figure 4.04. Diagram illustrating the average depths of each site on patch reefs located on different sides (aspects) and at varying distances from shore. Two replicate reefs were sampled in each zone across all depths and on all sides. The depths of the reefs in zone 1 completely overlapped. S: Southern aspect; T: Top of reef; N: Northern aspect. C = Central replicate; E: Eastern replicate.

193

Similarity in species distributions across sites on each reef A Bray-Curtis similarity matrix was generated for all observed coral species, based on their distributions across all reef sites. The results of the matrix was graphically displayed using both a dendrogram and a multidimensional scaling (MDS) diagram (Figs. 2.05, 2.06). Species that shared phylogeny and functional group membership tended to cluster together, both when plotted in a dendrogram and in multidimensionally scaled space. Coral species that fit this pattern included Madracis mirabilis and Madracis decactis; Diploria strigosa and Diploria labyrinthiformis; and Montastraea franksii and cavernosa. Species that did not fit the pattern of phylogenetic nor functional-group clustering are Porites astreoides [MV] grouping with Montastraea faveolata [MO], and the three species group of Favia fragum [MO], Siderastrea radians [MO] and Stephanocoenia intercepta [MV]. The two species Agaricia fragilis and Porites porites did not cluster with any species. Analysis of similarity (ANOSIM; Table 2.02) determined that the BV and MO functional groups were not significantly dissimilar in their distribution patterns in multidimensional space, but only marginally so (p = 0.06). The clusters formed by the MO and MV functional groups were not significantly dissimilar, at p = 0.1.

194

Figure 4.05. Dendrogram of Bray-Curtis similarities of species and functional groups clustered according by group-averaging. Similarities per pair of species were based on square-root transformed relative abundance data averaged across sites located on replicate reefs and over different aspects, depths and distances from shore.

195

Figure 4.06. Multidimensional scaling diagram (MDS) of square-root transformed relative abundance data for coral species averaged across sites located on replicate reefs and over different aspects, depths and distances from shore. The symbol used for each species indicates its functional group membership.

196

Table 4.02. Results of an ANOSIM analysis of the distinctness in clustering of each functional group. One-Way Analysis

Global Test Sample statistic (Global R): 0.395 Significance level of sample statistic: 2.5% Number of permutations: 999 (Random sample from 120120) Number of permuted statistics greater than or equal to Global R: 24

Pairwise Tests R Groups

Statistic

Significance Level %

Possible

Actual

Number >=

Permutations

Permutations

Observed

MV, PV

0.111

50.0

4

4

2

MV, MO

0.290

10.7

84

84

9

MV, BV

-0.222

100.0

10

10

10

PV, MO

1.000

14.3

7

7

1

PV, BV

-0.333

100.0

4

4

4

MO, BV

0.420

6.0

84

197

84

5

Similarity among sites in species assemblages The 60 sites surveyed across aspects, depths and zones were grouped according to how similar they were in terms of shared species, based on the relative abundance data. Sites were compared in order to see whether sites that shared aspect, depth or zone also shared species assemblage structure. The resultant matrix of Bray-Curtis site similarities, based on square-root-transformed relative abundance data for species, was graphed into both a dendrogram (not shown due to its complexity) and MDS. Linear boundaries of equal similarity were produced, enclosing similar sites within the two-dimensional graphic of multidimensional species state-space. Only the 58% iso-similarity boundary is illustrated, for clarity and because the site clusters generated by this one iso-similarity boundary were the most meaningful, as determined below. In order to determine whether there were meaningful patterns in the manner that the species relative abundance data clustered sites, three separate one-way and two-way analyses of similarity (ANOSIM) of sites were carried out. Sites were compared according to: 1. Aspect within depth and zone 2. Depth 3. Zone Sites were also graphed onto the three separate MDS diagrams using symbols that represented either aspect, depth or zone of each site. These three graphs are illustrated below.

198

1) Aspect Figure 4.07 illustrates the MDS of sites illustrated with icons that indicate Aspect. All of the sites located on the Tops of patch reefs (grey squares) are grouped in the upper cluster (A), along with sites with other aspects (the two triangle icons). Conversely, only sites that had either a Northern or Southern aspect are within the lower two clusters (B and C). There are no apparent clusters of northern vs. southern-facing sites. A series of Analyses of Similarity (ANOSIMs) of the factor Aspect across each depth within each zone confirmed this visual interpretation of the data. The tops of patch reefs possessed significantly different assemblages of coral species than either side of the same reef. The north and south sides of each patch reef, however, were only found to differ significantly in assemblage structure in Zone 1 and Zone 3 (Table 4.03 this table does not show the differences existed between sides in Zone 1 and 3 Table 4.04 shows that Zones 1 2 and 3D had differences between the sides I think the lack of a consistent pattern of differences will allow you to pool the north and south data The finding that coral assemblages were not dissimilar on opposite sides of patch reefs across most zones was unexpected. An analysis comparing the amount of light reaching surfaces with a southern or northern aspect found that significantly more light should be available to corals on the southern side of a patch reef in Bermuda than on the northern side of the same reef. It may be that the availability of diffuse light is sufficient to prevent the occurrence of differences between the two sides of patch reefs in most zones. On the other hand, it may be that the lagoonal corals are obtaining the majority of their energy via heterotrophy instead of autotrophy, and thus the coral assemblages on each site are not structured by light availability or the zoox are sufficiently adapted (more

199

pigments per cell) or more abundant (cells per mL or cm2 of coral tissue) that the coral is not penalized for the reduced light levels. Just have to be above saturating light levels, which may be only about 1/3- ½ incoming radiance for some zoox. Get some refs on this. As a third alternative, perhaps the metric of relative abundance that was used resulted in data that were not powerful enough (you needed more transects per “site”). for the detection of differences in the assemblage structure between the north and southern flanks of reefs. However, since tops of reefs in each zone were consistently found to differ significantly from that of both flanks but you had 8 transects compared to 4 on either flank, it seems likely that the relative abundance data should provide sufficient power for the complete analysis.

200

Figure 4.07. MDS of square-root transformed relative abundance data of species for all sites. Sites are represented by one of three icons depending upon the aspect which characterized that site. S: Southern aspect; T: Top of reef (i.e. vertical aspect); N: Northern aspect.

201

Table 4.03. ANOSIM table for the factor Aspect across all sites, including the results of pairwise post-hoc tests.

Global Test for Aspect Sample statistic (Global R): 0.098 Significance level of sample statistic: 0.1% Number of permutations: 999 (Random sample from a large number) Number of permuted statistics greater than or equal to Global R: 0

Pairwise Tests R Groups

Statistic

Significance Level %

Possible

Actual

Number >=

Permutations Permutations Observed

S, T

0.181

0.1

Very large

999

0

S, N

-0.001

43.6

Very large

999

435

T, N

0.198

0.1

Very large

999

0

202

Table 4.04. Significance levels of separate ANOSIM tests comparing similarities between reef sites located on the south versus north sides of reefs in each zone and at different depths. Sites on the sides of reefs are labeled as follows: S: Shallow; M: Mid-depth; D: Deep.

Zone and Depth 1 2 3S 3D 4S 4D 5S 5D 6S 6M 6D

p 0.009 0.046 0.591 0.008 0.103 0.349 0.263 0.288 0.119 0.417 0.834

203

2) Depth Figure 4.08 shows the MDS of sites illustrated with icons that indicate Depth. An obvious pattern can be seen in which sites are sorted by depth in the MDS diagram, (which is derived solely from species abundance data). ANOSIM revealed that globally sites clustered to a significant manner when categorized according to the factor Depth Pairwise comparisons indicate that sites categorized by virtually all depths were significantly clustered (Table 4.06). Only the species assemblages at depths 2 to 4 m, and 5 to 7m depths were not grouped significantly (Table 4.05). Light and wave energy both decline with water depth, and these two physical factors are probably responsible for the strong zonation patterns across depths in the coral species.

204

Figure 4.08. MDS of square-root transformed relative abundance data of species for all sites depths on all reefs. Sites are represented by one of ten icons that indicate the average depth in meters of that site. Darker icons represent deeper sites. The number in the legend next to each icon is the average depth in meters of that site.

205

Table 4.05. ANOSIM table for the factor Depth, calculated from data at each depth from across all sites.

Global Test for Depth Sample statistic (Global R): 0.382 Significance level of sample statistic: 0.1% Number of permutations: 999 (Random sample from a large number) Number of permuted statistics greater than or equal to Global R: 0

Table 4.06. Significance levels of pair-wise tests comparing similarities between reef sites located on different depths.

Pairwise tests: Significance Level 2 3 4 5

6

2 0.00 3

2 0.32

4

1

0.62 7

0.00 5

1

0.00 1

0.00 6

1 1

1 0.00

1 0.00

7

0.00 0.00 1

0.00 1

0.06 5

0.00 1

0.17 1

0.00 3

206

7

8

9

12

0.00 8

1

0.00 1

0.00 9

2 1

1

0.00 1

0.00

1

0.03 6

0.00

0.00

0.00

0.78 1 1 1 1 0.00 0.00 0.00 0.00 0.00

0.00 2

0.00

0.00

1 0.00

12

0.00

1

1

1

207

1

1

0.09 6

3) Zone There is also an obvious pattern in which sites are sorted consecutively by zone across the field of points in the MDS diagram (Figure 4.09), which is derived solely from species abundance data. Sites from Zones 1 and 2 are chiefly located in the lower right cluster, sites from Zone 3 and 4 are predominantly located in the upper cluster, and sites from Zones 5 and 6 are located in either the upper or lower left cluster. An ANOSIM of Zone (Table 4.07), regardless of aspect or depth of each site, determined that all zones were significantly different in terms of coral assemblage composition, except pairwise comparisons between Zones 1 and 2, and Zones 1 and 3. Zones 2 and 3 were significantly different, however. Table 4.08 displays the results of a SIMPER similarity analysis for each zone. SIMPER analysis examines the contribution of each species to the average resemblances between zones (Clarke et al. 2006). Generally it can be seen that the two Madracis species (which are in the Branched Viviparous functional group) dominate inshore zones, whereas the Massive Viviparous species Porites astreoides and the Massive Oviparous species Montastraea faveolata, Diploria strigosa and Diploria labyrinthiformis dominate offshore zones. One adaptive mechanism by which these five massive corals can coexist is examined in Section B, below.

208

Figure 4.09. MDS of square-root transformed relative abundance data of species for all sites. Sites are represented by one of six icons depending upon the zone that site was located within. Darker icons represent sites further from shore. The number in the legend next to each icon represents the zone the site was in, with Zone 1 being closest to shore and Zone 6 furthest from shore.

209

Table 4.07. ANOSIM table for the factor Zone across all sites.

Global Test Sample statistic (Global R): 0.622 Significance level of sample statistic: 0.1% Number of permutations: 999 (Random sample from a large number) Number of permuted statistics greater than or equal to Global R: 0

Pairwise Tests R Groups

Significance

Possible

Actual

Number >=

Statistic

Level %

Permutations

Permutations

Observed

1, 2

1.000

10.0

10

10

1

1, 3

1.000

6.7

15

15

1

1, 4

0.818

4.8

21

21

1

1, 5

0.929

2.8

36

36

1

1, 6

0.833

3.6

28

28

1

2, 3

0.722

2.9

35

35

1

2, 4

0.549

1.8

56

56

1

2, 5

0.895

0.8

120

120

1

2, 6

0.821

1.2

84

84

1

3, 4

0.35

4.8

126

126

6

3, 5

0.763

0.3

330

330

1

210

3, 6

0.792

0.5

210

210

1

4, 5

0.589

0.1

792

792

1

4, 6

0.617

0.2

462

462

1

5, 6

0.522

0.1

1716

999

0

211

Table 4.08. SIMPER analysis of the dominant species that differ between zones across the Bermuda Platform.

Species MDEC MMIR SRAD FAV MFAV MCAV

Zone 1 Average similarity: 71.57 Av.Abund Av.Sim Sim/SD 2.65 25.70 3.73 2.13 17.96 3.49 0.99 8.73 10.06 1.12 7.07 2.59 0.78 4.61 1.02 0.66 2.13 0.80

Contrib% 35.91 25.10 12.19 9.88 6.43 2.98

Cum.% 35.91 61.01 73.20 83.08 89.52 92.49

Zone 2 Average similarity: 68.78 Species Av.Abund Av.Sim MMIR 3.38 20.96 MDEC 2.47 17.48 PAST 1.82 10.65 MFAV 1.48 8.03 DSTRIG 0.77 4.01 MCAV 0.75 2.48

Sim/SD 2.13 2.93 3.72 2.47 1.48 0.76

Contrib% 30.48 25.41 15.49 11.68 5.83 3.60

Cum.% 30.48 55.89 71.37 83.05 88.88 92.49

Zone 3 Average similarity: 84.82 Species Av.Abund Av.Sim MMIR 3.79 24.55 MDEC 2.79 17.20 PAST 2.91 16.80 MFAV 2.30 9.98 MCAV 1.27 5.74 AFRAG 0.67 4.74

Sim/SD 10.50 3.31 3.30 2.72 158.50 1.80

Contrib% 28.94 20.28 19.80 11.76 6.77 5.59

Cum.% 28.94 49.22 69.02 80.78 87.55 93.14

Zone 4 Average similarity: 81.88 Species Av.Abund Av.Sim MMIR 2.94 17.31 PAST 2.78 14.40 MDEC 2.47 12.86 MFAV 2.49 12.11 DSTRIG 1.65 7.92 MCAV 1.23 6.43 MFRANK 1.12 4.18

Sim/SD 5.97 3.28 2.11 3.29 2.75 3.17 1.39

Contrib% 21.14 17.59 15.70 14.79 9.68 7.85 5.10

Cum.% 21.14 38.73 54.43 69.22 78.90 86.75 91.85

212

Table 4.08 continued. Zone 5 Average similarity: 74.30 Species Av.Abund Av.Sim PAST 2.17 13.18 MFAV 2.05 13.01 MDEC 2.25 12.37 MCAV 1.58 6.89 MFRANK 1.37 6.68 DSTRIG 1.13 6.42 DLAB 0.98 5.91 MMIR 0.86 4.16

Sim/SD 1.50 7.51 1.55 1.86 1.71 1.23 1.58 1.17

Contrib% 17.74 17.51 16.65 9.27 8.99 8.64 7.96 5.60

Cum.% 17.74 35.25 51.90 61.18 70.16 78.81 86.77 92.37

Zone 6 Average similarity: 73.62 Species Av.Abund Av.Sim PAST 2.01 14.07 MDEC 1.31 11.11 MCAV 1.28 10.00 MFAV 1.58 9.43 MFRANK 1.37 7.82 DLAB 1.06 6.16 DSTRIG 0.95 5.46 STEPH 0.65 4.41

Sim/SD 3.17 6.28 1.99 2.27 1.17 1.55 1.16 1.04

Contrib% 19.11 15.09 13.58 12.81 10.63 8.37 7.41 5.99

Cum.% 19.11 34.20 47.78 60.59 71.22 79.59 87.00 92.99

213

Aspect, Zone and Depth The aspect, depth and zone of a site interact to generate the cluster pattern Figs 4.07 – 4.09). The tops of sites from all zones are in the upper cluster (Figure 4.10). The sites located on the sides of reefs in Zones 1 through 4 are located in the lower right cluster (Figure 4.10), and sites located on the deeper sides of reefs in Zones 5 and 6 are located in the lower left cluster (Figure 4.10). Light and wave energy is higher on the tops of the lagoonal reefs, while deeper sites are darker and have slower current flow. Reefs nearshore have higher sediment than reefs offshore. Thus B sites are darker but with higher suspended sediment than A, and C sites are darker, deeper and with less suspended sediment than both A and B.

214

Figure 4.10. The sites surveyed across the Bermuda platform cluster into three groups with different coral species composition, which also match three different environmental conditions. The three groups are (A) Tops of all reefs, (B) deeper sites in zones 1-4 and (C) deeper sites in zones 5-6.

215

Distribution patterns of coral species Frequency of occurrence All corals as a group The average frequency of occurrence (RF) of any coral at each site is plotted as dot size onto the same MDS graph used above (Figure 4.11) The same data was also plotted using the standard format for a line graph (Figure 4.12A) and also as a “bubble” graph (Figure 4.12B). These three graphs allow one to examine how corals overall were distributed across the patch reefs located over the lagoon. All three graphs illustrate that corals occurred frequently on many shallow sites across most zones. The smaller sized circles in the lower left cluster indicate that corals occurred less frequently at sites that were located on the sides of patch reefs far from shore.

216

Figure 4.11. MDS of square-root transformed frequency of occurrence data of all coral species for all sites, as in the three figures above. In this and all following MDS diagrams throughout this section, the size of each circle represents the number of frames containing any coral, averaged across transects per site, according to the scale in the legend.

217

A,

B.

218

Figure 4.12. The average proportion of frames with any coral present across all sites, illustrated as a line graph per site per reef (A) and as a bubble graph per depth and zone (B). Branched Viviparous Corals The frequency of occurrence of all branching viviparous corals as a group, and of each species of BV coral separately, are plotted in Figs 4.13 to 4.17 below. As a group, the BV corals occur most frequently on the sides of the reefs in zones 1 and 2, and occur less often on reefs further from shore. The two species of Madracis each separately share this basic distribution pattern, with a much greater likelihood of occurring in frames sampled on the sides of patch reefs rather than on the tops, and greater numbers of occurrences observed nearshore compared with offshore (Figs 4.13 B, 4.13C, 4.15, 4.16). In comparison, the species Porites porites differs from the Madracis species in its distribution, and was only observed to occur on the tops of reefs and never on the sides. Porites porites was also much rarer than the Madracis species across all reef sites overall (Figs. 4.13D, 4.17).

219

Figure 4.13. Four MDS graphs of square-root transformed frequency of occurrence data of Branched Viviparous species as a group, and for M. decactis, M. mirabilis and P. porites corals separately for all sites.

220

Branched Viviparous Functional Group A.

B.

221

Figure 4.14. The average proportion of frames with corals of the Viviparous Branching (VB) functional group present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom). Madracis decactis A.

B.

222

Figure 4.15. The average proportion of frames with corals of the species Madracis decactis present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom). Madracis mirabilis A.

223

B.

Figure 4.16. The average proportion of frames with corals of the species Madracis mirabilis present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom). Porites porites A.

224

B.

Figure 4.17. The average proportion of frames with corals of the species Porites porites present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom).

225

Foliose and Plating Viviparous Corals The only coral in the foliose and plating (FP) functional group that was observed within the lagoon was the species Agaricia fragilis. The distribution of A. fragilis was found to have little similarity to the distribution of the other corals assessed using BrayCurtis similarity analysis (Figs 4.18, 4.19). A. fragilis was observed primarily on the deeper flanks of patch reefs in Zones 1–3. Its predominance in low light environments fits the prediction for A. fragilis as a stress-tolerant species adapted to areas of low current flow, low light availability and also low total coral cover.

226

Figure 4.18. MDS graph of square-root transformed frequency of occurrence data of the coral species Agaricia fragilis, the only member of the Foliose and Plating Viviparous (FP) functional group observed in the North lagoon in Bermuda in this study. Circles size represents the average number of frames containing A. fragilis per site out of 20, as in the legend on the side of each MDS graph. Note the different scale used in this MDS for A. fragilis compared with the other MDS diagrams in this section.

227

Agaricia fragilis A.

B.

228

Figure 4.19. The average occurrence frequency of the species Agaricia fragilis across all sites, illustrated as a line graph per site and reef (top) and as a bubble graph per depth and zone (bottom). Massive Viviparous corals When the frequency of occurrence of all species of the Massive Viviparous functional group, and each member species, Favia fragum, Porites astreoides and Siderastrea radians, are plotted at each site across the MDS diagram (Figure 4.20) we can see that the distribution pattern of the MV group (Figure 4.20) is primarily driven by the distribution of P. astreoides (Figure 4.20C; 4.22). P. astreoides is most abundant in the upper cluster, which represents the tops and shallow sides of reefs. Sites in the lower two clusters, which are derived from reef sites located on the deeper sides of patch reefs, are characterized by much less P. astreoides. Favia fragum is less abundant and is also primarily found on the tops of reefs (Figs. 4.20B; 4.22). Alternatively, S. radians appears equally distributed across reef sites across the three clusters (Figure 4.20D; 4.24), with no obvious pattern related to depth, zonation or aspect.

229

Figure 4.20. Four MDS graphs of square-root transformed data of occurrence frequency by Massive Viviparous species as a group, and for F. fragum, P. astreoides and S. radians corals separately, for all sites. Circle size represents the average number of frames containing each species or functional group per transect per site out of 20, as in the legend on the side of each MDS graph.

230

Massive Viviparous functional group A.

B.

Figure 4.21. The average proportion of occupied frames per transect with corals belonging to the Massive Viviparous functional group present across all sites,

231

illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom). Favia fragum A.

B.

232

Figure 4.22. The average proportion of frames with corals of the species Favia fragum present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom).

233

Porites astreoides A.

B.

234

Figure 4.23. The average proportion of frames with corals of the species Favia fragum present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom). Siderastrea radians A.

B.

235

Figure 4.24. The average proportion of frames with corals of the species Siderasterea radians present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom). Massive Oviparous corals As a group, massive oviparous corals were most abundant on the tops of reefs (upper cluster) and on the sides of the reefs in Zones 5 and 6 (left cluster; Figs. 4.25A, 4.26). Only the species M. faveolata (Figs. 4.25C, 4.27) displayed a similar distribution pattern to the pattern displayed by the functional group overall, however. The two species M. cavernosa and M. franksi, which were found to be similar at a 75% level using BrayCurtis analysis, did display obvious differences in distribution when graphed. M. cavernosa was most abundant on the sides of nearshore reefs (right cluster) and on the tops of reefs (top cluster; Figs. 4.25B, 4.26), while M. franksi was most abundant on the sides of the offshore reefs (Cluster C; Figs 4.25D, 4.28). The two Diploria species exhibit similar patterns of abundance (Figs 4.25E, 4.25F, 4.29, 4.30), with most corals occurring on the tops of reefs in the upper cluster, and few species on the flanks in the other two habitat clusters. Some differences in distribution between the two Diploria species are apparent, however. D. labyrinthiformis is less abundant than D. strigosa overall, and appeared to be more common on the sides of reefs offshore (left cluster) relative to D. strigosa. Stephanocoenia intersepta displayed the most distinct pattern of abundance relative to the other MO corals (Figs. 4.25G; 4.31). Its highest abundance was not on the tops of reefs in Zones 3 and 4 like most MO coral species, but instead on the sides of reefs in zones 5 and 6.

236

Figure 4.25 . Seven MDS graphs of square-root transformed relative abundance data of (a) Massive Oviparous species as a group, and for (b) M. cavernosa, (c) M. faveolata, (d) M. frankesi, (e) D. labyrinthiformis, (f) D. stigosa and (g) S. intersepta separately for all sites.

237

Massive Oviparous functional group A.

B.

238

Figure 4.26. The average proportion of frames with corals of the Massive Oviparous (MO) functional group present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom).

239

Montastraea cavernosa A.

B.

240

Figure 4.27. The average proportion of frames with corals of the species Montastraea cavernosa present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom). Montastraea faveolata A.

B.

241

Figure 4.28. The average proportion of frames with corals of the species Montastraea faveolata present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom). Montastraea franksi A.

242

B.

Figure 4.29. The average proportion of frames with corals of the species Montastraea franksi present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom). Diploria labyrinthiformis A.

243

B.

Figure 4.30. The average proportion of frames with corals of the species Diploria labyrithiformis present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom). Diploria strigosa

244

A.

B.

Figure 4.31. The average proportion of frames with corals of the species Diploria strigosa present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom).

245

Stephanocoenia intersepta A.

B.

246

Figure 4.32. The average proportion of frames with corals of the species Stephanocoenia intersepta present across all sites, illustrated as a line graph per site per reef (top) and as a bubble graph per depth and zone (bottom).

Standard measures for coral reefs In the section below I examine how average percent cover of the entire coral assemblage, and each functional group separately, as well as species richness and functional group richness vary over the six zones. Due to time constraints only the tops of three replicate reefs in each zone were surveyed. The data from these three surveys are presented first. On two of the three replicate reefs, both the tops and south sides were surveyed for the six factors. The data that includes the change in coral assemblages across the sides of patch reefs are described in the section following the one below.

Section 1: Tops of Reefs Only A. Average Percent Coral Cover The average percent coral cover for all corals as a group (which I term “Total Coral Cover” or TCC below) peaked in the middle of the north lagoon in Zone 3 (Figure 4.33A). TCC appeared substantially lower in Zone 1, nearshore. TTC also declined progressively with distance from shore across Zones 4 to 6. Additionally there appeared to be substantial variability in TCC among reefs within each region. Two-way ANOVA confirmed that the interaction between zone placement and replicate reefs was highly significant, at p < 0.001 (Table 4.09A). Since wave energy has been shown to be higher offshore (Mills et al. 2004), and suspended sediment load higher nearshore (Toro Farmer,

247

unpublished data) it may be that the central reefs have higher TCC because the corals there are exposed to less levels of disturbance or stress than corals on reefs either offshore or nearshore.

B. Species Richness The number of coral species on the top of reefs peaked in Zone 2 (Figure 4.33B). The variability in species richness between sites was highest in Zone 1 and declined on sites further from shore. Two-way ANOVA confirmed that the interaction of zone placement and replicate was highly significant, at p = 0.006 (Table 4.09B). This pattern, in which nearshore sites display higher variability than offshore sites, was hypothesized by Murdoch and Aronson (1999), and reiterated by Pandolfi (2002). Zone 2 represents an area in close proximity to the south shipping channel, as well as near the island of Bermuda. It may be that disturbances are more frequent in this area and that the mechanisms of the Intermediate Disturbance Hypothesis (Grime 1977; Connell 1978) are operating to promote coral richness there.

C. Functional Group Richness Functional group richness (Figure 4.33C)for the three predominant functional groups was lowest offshore, reached a plateau across sites 2–4, and then declined again in Zone 1. Two-way ANOVA confirmed that the interaction of zone placement and replicate was highly significant, at p = 0.005 (Table 4.09C).The analysis of each species in the section above and for percent cover of each functional group illustrates that the decline in FG richness offshore is due to a decline in the abundance and biomass of the Branched

248

Viviparous functional group. The decline in Zone 1, alternatively, is due to a decline in the abundance and biomass of the two mound-shaped or massive functional groups.

D. Percent cover of the Branched Viviparous Functional Group Branched viviparous corals are predicted to be both competitive and ruderal (C-R), following the AST model of Grime (1977). As such they should be capable of tolerating disturbance to at least a moderate degree, and also capable of dominating habitats when released from competition (See Ch. 1). Branched viviparous corals peaked in abundance on some of the reefs in Zone 2 in this analysis (Figure 4.33D). Sites in all other zones exhibited substantially lower cover of the BV functional group, with virtually zero cover in Zones 5 and 6. Two-way ANOVA (Table 4.09D) confirmed that the interaction of zone placement and replicate was highly significant, at p < 0.001. Zone 2 is nearest the southern shipping channel, which is a source of suspended sediment during summer months. Others have noted that the branched corals dominate nearshore habitats off North Shore (i.e. Logan 1988; Mills et al. 2004). It may be that the BV functional group is more tolerant of suspended sediment than the massive corals in the other functional groups, and that this tolerance, combined with a lack of competition with the other groups allows the branched viviparous corals to dominate reefs in Zone 2.

E. Percent cover of the Massive Viviparous functional group The massive viviparous corals are predicted to be the most tolerant of disturbance. As such the cover of the MV group was expected to show little variation across sites regardless of zone. As illustrated in Figure 4.33E, however, the average percent cover of

249

this functional group peaked in Zone 4, with lower cover values in zones nearer or further from shore. Two-way ANOVA confirmed that the interaction of zone placement and replicate was highly significant, at p < 0.001 for the MV functional group (Table 4.09E). F. Percent cover of the Massive Oviparous functional group The massive oviparous corals are predicted to be relatively tolerant to stress but also to be able to dominate reefs through competitive interactions when resources are abundant and disturbance levels are low. The MO functional group is predicted to display low cover on reefs exposed to higher disturbance. In the analysis of percent cover on the three reefs across the six zones across the North Lagoon of Bermuda, the MO functional group displayed similarly high values across reefs in zones 3 through 6, with a decline in cover on reefs in Zones 2 and 1 (Figure 4.33F). Variability between reefs within zones was minimal in Zone 4, with more variability evident in zones closer or further from shore. Two-way ANOVA (Table 4.09F) confirmed that the interaction of zone placement and replicate was highly significant, at p < 0.001. It may be that the higher levels of turbidity and lower current flow speeds nearshore caused a decline in survivorship for the large mound-shaped corals in the MO functional group. Reefs offshore, however, are exposed to much lower levels of sediment, but higher wave energy. The MO corals should be able to withstand wave energy due to their morphology, so it is unclear why the percent cover of the MV functional group would not be highest on reefs furthest from shore

250

Figure 4.33. (next page). Average percent coral cover for (A), (B) species richness, (C) functional group richness as well as (D – F) the average percent cover for each functional group on the tops of each of three replicate reef sites in each of the six zones. The data for each site was based on eight transects of 10-m length. The tops of the reefs of each zone differed in depth, and light availability at a given depth was reduced nearshore relative to offshore, as described above.

251

A. Total Coral Cover D. Branched Viviparous Coral Cover

B. Species Richness E. Massive Viviparous Coral Cover

C. Functional Group Richness F. Massive Oviparous Coral Cover

252

Table 4.09. Results of the 2-way ANOVAs of the six parameters across the 18 reef sites located on the tops of patch reefs located in three replicate “legs” across six zones located across the north lagoon in Bermuda. A. Total Coral Cover Source SS ZONE 1.760 LEG 0.023 ZONE*LEG 0.612 Error 0.878

df 5 2 10 126

MS 0.352 0.012 0.061 0.007

F-ratio 50.501 1.670 8.779

B. Species Richness Source SS ZONE 109.285 LEG 29.056 ZONE*LEG 36.111 Error 173.875

df 5 2 10 126

MS 21.857 14.528 3.611 1.380

F-ratio 15.839 10.528 2.617

p-value 0.000 0.000 0.006

C. FG richness Source ZONE LEG ZONE*LEG Error

df 5 2 10 126

MS 3.990 0.361 0.586 0.215

F-ratio 18.535 1.677 2.723

p-value 0.000 0.191 0.005

D. Branched Viviparous Coral Cover Source SS df ZONE 1.518 5 LEG 0.065 2 ZONE*LEG 0.640 10 Error 0.647 126

MS 0.304 0.032 0.064 0.005

F-ratio 59.148 6.320 12.466

p-value 0.000 0.002 0.000

E. Massive Viviparous Cover Source SS ZONE 0.788 LEG 0.003 ZONE*LEG 0.240 Error 0.341

df 5 2 10 126

MS 0.158 0.001 0.024 0.003

F-ratio 58.255 0.553 8.860

p-value 0.000 0.576 0.000

F. Massive Oviparous Cover Source SS ZONE 1.249 LEG 0.197 ZONE*LEG 0.544 Error 1.062

df 5 2 10 126

MS 0.250 0.098 0.054 0.008

F-ratio 29.632 11.663 6.451

p-value 0.000 0.000 0.000

SS 19.951 0.722 5.861 27.125

p-value 0.000 0.192 0.000

254 Standard measures for coral reefs Section 2: Tops and South Sides of Reefs A. Average Percent Coral Cover As in the section above, the average percent cover of the entire coral assemblage (i.e. TCC; Figure 4.34A) can be seen to vary substantially both across zones and across replicates within zones. Depth also clearly plays a factor in the total coral cover at a site, and appears to interact in a complex manner with distance from shore. Coral reefs nearshore in Zones 1 to 3 display higher cover on the sides of reefs than on the tops of reefs, while reefs in the zones further from shore display higher cover on the tops of reefs than on the sides. TCC is highest on the flank of one of the 2 reefs samples in Zone 2. TCC was found to be lowest on the top of reefs in Zone 1, and at the deepest sites on the reefs in Zone 6. Historically researchers in Bermuda have only surveyed the tops of patch reefs found in the lagoon. As can be seen by comparing Figs. 4.34 and 4.35 to Figure 4.33 from the last section, surveying only the tops of reefs fails to account for a substantial proportion of the average percent cover that occurs on these lagoonal reefs. Instead, to truly represent the condition of Bermuda’s lagoonal reefs, surveys should encompass the full range of depths that occur across each reef, and over the entire extent of the lagoon.

B. Species Richness On the reefs surveyed in this project, species richness (Figure 4.34B) can be seen to vary with depth as well as zone and across replicates within zones in a complicated manner. Species richness appeared to peak in Zone 3, was found to be lowest on the

255 deepest sites in Zone 6. Surveys that only encompass the tops of reefs, as in Section 2 above, fail to uncover these complexities in species richness that occur across the flanks of lagoonal reefs in Bermuda.

C. Functional Group Richness The number of functional groups within a site also appears (Figure 4.34C) to be due to factors that interact across zones, replicates and depths. Functional group richness appears higher on deeper sites on reefs nearshore than on the tops of reefs. Alternatively, functional group richness is lower on the tops of reefs offshore than it is on the sides of the same reefs. Functional group richness peaks in Zones 3 and 4, and is lowest on the deepest sites in Zone 6. Again, ecologically important patterns in functional group richness are apparent on the sides of these lagoonal patch reefs that standard survey techniques would have missed.

D. Percent cover of the Branched Viviparous functional group The percent cover of the coral species belonging to the Branched Viviparous functional group was found to peak in Zone 2 (Figure 4.35A). Moreover, percent cover of this group was consistently higher on the deeper sites than on the tops of reefs in Zone 1 through to Zone 4. The BV functional group displayed low percent cover in Zones 5 and 6 across all depths. Since the BV corals were consistently higher on the sides of patch reefs than on the tops, it is apparent that standard surveys which only consider the tops of patch reefs will misrepresent the contribution of the BV coral species to the assemblage structure of lagoonal corals.

256 When the percent cover of each of the member species of the BV functional group are presented separately (Figure 4.36), it is apparent that the species Madracis mirabilis contributed the greatest percentage of biomass and thus was responsible for the general pattern in percent cover that characterized the functional group as a whole. Madracis mirabilis peaked in Zones 2 with a decline in cover both closer and further from shore. It also displayed higher cover on the flanks of patch reefs versus on the tops. The sibling species to Madracis mirabilis, Madracis decactis, displayed a similar distribution and pattern of biomass, but with lower values overall (Figure 4.36B). Conversely the third BV coral, Porites porites (Figure 4.36C), displayed a completely different pattern in terms of percent coral cover. It was found to exhibit the highest percent cover values on the tops of reefs instead of the flanks like the other BV species. P. porites also displayed very low values in percent cover, generally being represented by only one point per site. The species-specific patterns apparent in the percent cover data of these BV coral species match the patterns displayed by the same species in the MDS analysis described above, which were based on measures of relative abundance data instead.

E. Percent cover of the Massive Viviparous functional group The coral species of the Massive Viviparous functional group consistently peaked in cover on the tops of reefs and was lower on the flanks of the same reefs, regardless of zone (Figure 4.35). The percent cover of the MV functional group peaked in Zone 3, and was found to be lower on reefs closer to shore and further from shore. Surveying the tops of reefs for this functional group would provide meaningful estimates of its functional role across the lagoon.

257 Comparison of the component species of the MV functional group (Figure 4.38) illustrates that the species P. astreoides is responsible for the general pattern in percent cover across the lagoon that characterizes the group. P. astreoides peaks in abundance in Zones 3 and 4, and exhibits higher percent cover on the tops of reefs versus on the deeper flanks. Favia fragum and Siderastrea radians, alternatively, present very low percent cover values of 1 or 2 points per site, and appear to occur across depths and zones. These patterns concur with the patterns apparent in the MDS analysis based on relative abundance data described in Section 1 above.

F. Percent cover of the Massive Viviparous functional group The corals that comprise the Massive Viviparous functional group were found to peak in percent cover on the tops of reefs in Zones 3 to 6, with lower values apparent in Zones 1 and 2, nearer to shore (Figure 4.35). The cover of the MV functional group generally declined with depth across all zones, but with an alternate pattern occurring on one of the two replicate reefs in some zones. The two Diploria species peaked in percent cover on the tops of patch reefs and displayed lower cover on the sides, across all zones (Figure 4.38). The species Diploria strigosa exhibited higher cover than D. labyrinthiformis, although both species peaked at roughly 5% cover on the lagoonal reefs. These co-occurrence patterns of these two species were found to be very similar, based on relative abundance data (section 1 above), and the percent cover data appears to confirm that these two sister species also share habitats when biomass is used as a measure.

258 The percent cover data for the two species Montastraea frankesi and M. faveolata was grouped into one variable after it was decided that the discrimination between the two species in the point count analysis may have been compromised. The M. annularis spp. category displayed the highest percent cover across reefs, with high variability in cover apparent across depths, replicate reefs and zones. The congener M. cavernosa was found to contribute lower percent cover across the region overall. It peaked in cover on the sides of patch reefs in the middle of the lagoon, with lower values on reefs in Zones 1, 2 and 6. The species Stephanocoenia intersepta was rarely sampled by point count analysis. The species was absent from the two sampled reefs in Zone 4, and it peaked in cover in Zones 1 and 6. It also tended to display higher percent cover on deeper sites on a reef than on the tops of reefs. The pattern of cover presented by the species Stephanocoenia intercepta appears substantially different to that of the other species of the MO functional group. The patterns generated by the percent cover data for all species in the MO functional group matches the patterns determined by relative abundance data as described in Section 1 above.

G. Percent cover of the Foliose and Plating Viviparous functional group The one species of the FP functional group that was sampled across the lagoon was the plating species Agaricia fragilis (Figure 4.35). It was only represented in the percent cover data by a few points or less per site, and always on the deeper sites on inshore reefs located in Zones 1 to 3. Observations while diving found that, while not represented by point count data, the species A. fragilis was only observed on the tops of patch reefs in

259 Zones 1 and 2, and elsewhere generally within caves and in the shadows of overhanging ledges.

Figure 4.34. (Below) Percent cover, species richness and functional group richness of corals surveyed on sites located on the tops and southern flanks of patch reefs. Two replicate reefs were sampled across six zones. T: Tops; S: Sides. Numbers represent depths in feet [and will be converted to meters and make larger!].

Figure 4.35. (Below) Percent cover of the Branched Viviparous, Massive Viviparous and Massive Oviparous functional groups of corals surveyed on sites located on the tops and southern flanks of patch reefs. Two replicate reefs were sampled across six zones. T: Tops; S: Sides. Numbers represent depths in feet [and will be converted to meters].

260

261

262

Figure 4.36. Percent cover of the three species of Branched Viviparous functional group.

263

Figure 4.37. Percent cover of Agaricia fragilis, the one species of the Foliose and Plating functional group found within the lagoonal sites surveyed.

264

Figure 4.38. Percent cover of the three species of Massive Viviparous functional group.

265

Figure 4.39 A-B Percent cover of the two of the five species of Massive Oviparous functional group.

266

Figure 4.39 C-E. Percent cover of the three other species of Massive Oviparous functional group.

267

DISCUSSION

Species and functional group distributions across sites In the investigation regarding whether species that were members of the same functional groups shared distribution patterns in terms of occurrence, the MDS of coral species based on relative abundances across sites fit a pattern indicating nested functional groups across habitat types, (Figure 1.07). This mixed pattern was generated because some, but not all, species within each functional groups shared similar distribution patterns across sites. In each functional groups there was a group of species that shared habitats, with a small number of species displaying low coverage, relative abundance and no clear preference or occurrence within particular habitats. For instance, the BV coral Porites porites differed from the other species of Branched Viviparous corals in both habitat and biomass. In a similar manner, the MO corals Favia fragum and Siderastrea radians were most similar to the MV coral Stephanocoenia intersepta, in that all three species were relatively rare, exhibited low coverage , and did not show a clear preference for either the tops of reefs nor the deeper sites. Nor did these three species show a preference for reefs inshore nor offshore. Of the species within functional groups that did share distribution patterns, and thus high levels of Bray-Curtis similarity, most pairs were composed of congeneric species. This pattern in which species that share phylogenetic heritage also share habitats implies that the environmental tolerances of the sibling species act to control their distribution to a greater degree than intergeneric competition. One mechanism by which sister species

268 could co-occur within habitats include a priority effect in which each species has an equal likelihood of occurring within a small patch and each can prevent its congener from removing it (Brown et al. 2002) . Alternatively it may be that, since coral cover on many of the sites surveyed is less than 50%, competitive interactions are rare and the effects of competition weak across the lagoon and that environmental conditions limit the longevity of all colonies. If so then continual recruitment by all species would allow persistence across the lagoon despite short life spans among individuals of each species. Statistical tests that can check for the pattern in which species share habitats at a larger scale, while rarely sharing patches within habitats could be used to determine whether intergeneric competition operates in the species that showed shared distributions in the Bermuda lagoon. The occurrence of species within functional groups that possessed disparate distribution patterns represents a result that is contrary to the predictions of the AST (Grime 1979). Within all three functional groups there were some inter-group distributional differences between species that indicate that the AST theory has some limitations in its ability to predict species distributions. Within the MO functional group, the two Diploria species appeared to dominate the tops of reefs, while Montastraea species were abundant on both the tops and sides of reefs. Conversely Stephanocoenia intersepta was primarily found at the deepest parts of offshore reefs. Similarly the BV coral P. porites was located on the tops of reefs while the two Madracis species were found on the flanks of patch reefs. Porites astreoides also dominated the tops of reef while S. radians, another MV coral, typically occurred at the base of reefs. In all three functional groups it appears as if each functional group has member species that fulfill

269 their functional role within a particular depth, with different species representing each functional group at a different depth. Such a distribution pattern may be indicative of the role of inter-group competition in limiting the membership of species within a functional group to the same depth habitat, while shared physiological traits concurrently result in shared life-history traits and functional responses across depths by members of the same functional group.

Percent coral cover per functional group In the North Lagoon of Bermuda, different functional groups of corals varied in relative abundance, percent cover and in occurrence across a range of habitats in a manner that supports some of the hypotheses of the modifies AST presented in Chapter 1. The Massive Oviparous (MO) and Branched Viviparous (BV) functional groups were characterized in Ch. 1 as stress-tolerant–competitive and ruderal–competitive, respectively. As such both groups were predicted to dominate particular habitats in the absence of the BO functional group, which is predicted to be the most competitive, but which is not found in Bermuda. Alternatively, the Massive Viviaparous (MV) and FP (Foliose and Plating) functional groups were predicted to be ruderal and stress-tolerant, respectively. Both of these groups are predicted to not dominate habitats in terms of abundance or biomass, although for different functional reasons. The MV functional group is predicted to allocate resources to reproduction over biomass accumulation as a means of surviving the effects of competition and disturbance. Conversely the FP functional group is predicted to occur in habitats characterized by low levels of light where disturbances and competitive interactions are rare.

270 Management issues Previous research on cross-lagoonal patterns of coral species distributions has been limited to only four locales: the nearshore patch reefs, the outer lagoonal reefs at Crescent and Three Hill Shoals, the rim reefs around North Rock and the 30-ft forereef (Logan 1988; Smith et al. 2003; Jones 2007). The previous interpretations of the data from these locations paint the picture that coral cover and diversity increases in a fairly linear manner from nearshore to offshore. This simple gradient in biomass and diversity is hypothesized to be due to the negative influence of factors created by the island, such as sediment or temperature extremes. However, the previously surveyed areas are separated by large expanses of reef that have never been assessed scientifically. As such, the precision of this historical research is rather coarse, and more complex patterns of abundance, biomass and diversity may in fact exist across the North Lagoon. Since I surveyed replicate reefs over six consecutive zones across the lagoon, the intensive sampling represented by the current study should provide a more accurate depiction of the condition of the coral assemblages in question. In terms of functional effects, both the MO and BV functional groups provide the most coral cover across the Bermuda lagoon, although in different zones. All functional groups of corals possess a range of different characteristics, and management actions that promote the survivorship of one functional group may inhibit the survival of species belonging to other groups. In the same manner that botanists farm or garden trees in a different way than they tend grasses or herbs, coral reef managers and scientists need to consider the functional responses of each functional type of coral, and not treat all corals as functionally equivalent in response or effect.

271 While the species of the MV functional group does not provide much coral cover, its component species are highly fecund and are the most likely to recruit to newly created habitat, if water quality and other conditions are appropriate. The BV and MO corals recruit more rarely but may rely on similar cues as the MV functional group. As such the MV functional group may be a useful species for indicating the relative intensity of disturbance within a habitat, and also the relative suitability of the area for recruitment by species belonging to all functional groups of coral. The highest percent cover, diversity and relative abundance of corals was observed to be within Zones 3 and 4, which have been neglected by previous researchers. These reefs are located with the North Coral Reef Preserve and so are legally protected,. The reefs are also in close proximity to the south shipping channel, however, and as such are at a greater risk to environmental impact compared with most reefs in Bermuda. This project also determined that the coral assemblages found on the flanks of the lagoonal patch reefs possess higher coral cover and diversity than the tops of the reefs. Most coral research focuses on the tops of patch reefs, due to the historical focus on forereef sites in which flanks do not exist. Research and management of the lagoonal reefs in Bermuda, and presumably elsewhere, may be driven by erroneous data unless the sides of the reefs are considered. The deeper reef sites in Zone 6 were characterized by low coral abundance and percent cover, despite the proximity to the open ocean, the higher availability to light compared to sites at the same depth nearshore, and a lower range of temperature variability. The cause for the low biomass and species richness in this habitat is not

272 obvious, and may have been due to some form of disease or anthropogenic disturbance heretofore unrecorded.

273

274

CHAPTER 5: SUMMARY, CONCLUSIONS AND RECOMMENDATIONS

In this dissertation I modified the Adaptive Strategies Theory, originally developed by Grime (1977) for terrestrial plants, by removing confounding variables from the visual model. The original Adaptive Strategies Theory relied strongly on a ternary model showing the range of habitat characteristics across which species could adapt. However, this ternary model confounded the independent variables of resource availability and disturbance with the dependent variable of competition. I restructured the model as a twodimensional box across which only the two independent variables of resource availability and disturbance were plotted. By modifying the model in this way, I was better able to consider the varying levels of resource gain and loss, and thus use resource economics theory to predict adaptive strategies. I then illustrated how the necessary trade-offs required to cope with differing rates of resource gain and loss determine the range of physiological and behavioral responses exhibited by an organism across habitat types. I applied the refined theory to Caribbean reef corals. Corals were sorted into functional groups based on their morphologies and reproductive modes. These two functional attributes are indicative of the following adaptive strategies:

• Branched oviparous corals [BO]: Competitive dominant, • Branched, viviparous corals [BV]: Competitive-Ruderal,

275

• Massive, viviparous corals [MV]: Ruderal, • Massive, oviparous corals [MO]: Competitive – Stress-Tolerant, • Plating, foliose & solitary corals [FP], (which are only viviparous in the Caribbean): Stress-Tolerant In both Florida and Bermuda, these functional groups of corals responded to gradients of disturbance and stress in predictably different ways. In Florida, despite the chaotic patterns in biomass displayed by each assemblage of coral species when separately plotted across reefs, each functional group of corals responded to direct and indirect gradients of disturbance in a orderly and group-specific manner. The replacement pattern predicted by Grime (1977), in which each functional group dominated a particular region of the gradient, was not observed (Figure 5.01A). Instead, functional groups displayed a nested distributional pattern (Figs. 5.01B; 5.02; 5.03), indicating that negative interactions between functional groups are probably weak . In Bermuda as well, functional groups displayed a nested pattern across sites located over a range of depths and geomorphological reef zones (Figs 5.01B; 5.03). As predicted by the modified Adaptive Strategies theory for nested functional groups (Figs. 5.03; 5.04), branched viviparous corals dominating mid-depth sites inshore which were characterized by low disturbance by waves and high resource availability from light and suspended particulate matter. The massive viviparous corals, which were predicted to be ruderal and thus limited to high resource habitats regardless of disturbance level, primarily occurring in shallow sites from mid-shelf to offshore. The massive oviparous corals, which were predicted to be more stress-tolerant and thus occur across a greater

276 depth gradient, but a more limited disturbance gradient, dominating both shallow and deep sites from mid-shelf to offshore.

277

Figure 5.01. Diagrams depicting the differing ways in which the abundances of competitive (C), stress-tolerant (S) and ruderal (R) functional groups of corals are predicted to vary across habitats located across the range of stress and disturbance gradients encompassed by the AST model, and depending on the degree of niche overlap exhibited by each functional group. In all graphs X and Y represent graphs of abundance relative to levels of disturbance (X) or stress (Y). Z represents the modified CSR square diagram, with the zero net growth intercept (ZNGI) illustrated for each functional group. Graph A matches Grime’s original predictions in which functional groups are limited to specific regions of adaptive niche phase space with no overlap in niche boundaries. Graph B represents an alternate model in which competitive species do not negatively interact with stresstolerant or ruderal species. In graph B the functional groups exhibit maximal overlap in niche boundary. In all models C, S an R functional groups maintain dominance under differing environmental conditions.

278

Figure 5.02. A diagram illustrating how the Zero Net Growth Intercepts (ZNGI) of each of the predominant functional groups of Caribbean coral found in Florida and Bermuda are dispersed across the Adaptive Strategies Theory model. For clarity, the inset shows the distribution for each functional group separately. The functional groups are distributed in a nested pattern across habitat types defined by varying rates of resource gain and loss. Letters represent the functional group occupying each patch as described in the text above. The dark gray field on the lower right side of the diagram represents

279 the range of habitats in which high relative rates of resource loss limit biomass.

Figure 5.03. The bounded area laid over the modified Adaptive Strategies Theory shown in Figure 5.02 represents the range of habitat types surveyed in Florida.

280

Figure 5.04. The bounded area laid over the modified Adaptive Strategies Theory shown in Figure 5.02 represents the range of habitat types surveyed in Bermuda. Numbers represent zones, with 2 closer to shore and 6 furthest offshore. Letters represent relative depths, as follows: S: Shallow; D: Deep.

In Florida the functional-group approach provided new insights into the manner in which varying levels of disturbance affected species richness across sites. Massive

281 oviparous corals represented approximately 85% of the percent coverage of all corals on a reef regardless of overall coral cover at a site, whereas other kinds of coral represented a much smaller proportion of overall coral cover. Despite the small proportions in overall cover, there were dramatic changes in species presence patterns across sites by the subordinate functional groups. By categorizing species into functional groups, and then tabulating the presence or absence of species across reefs ranked according to water quality or overall coral cover, it was apparent that only the branched viviparous (BV) and the Foliose and Plating (FP) functional groups lost species across the gradient. The BV functional group lost species at both high and low total coral cover, whereas the FP functional group had progressively fewer species as coral cover declined across reefs. The loss of stress-tolerant species such as those in the FP group was predicted by the modified Adaptive Strategies Theory. When species were aggregated according to shared habitat in Bermuda, species from the same genus co-occurred in almost every case. This implies that these closely related species also share many functional traits and yet still coexist in many habitats. There are a number of strategies by which these closely related species may coexist, including: 1. The related species may have either evolved means of avoiding competition, are ruderal, or are stress tolerant and therefore are probably located within a (neutral; Hubbell 2001) habitat where competition is so slow as to be negligible. 2. The species compete heavily at the colony scale but are able to mitigate competition on a smaller or larger scale. 3. The species are so similar that at the population level there is competitive equilibrium.

282 4. The species have slight differences in the location of source and sink populations across the reef platform. Each of the above points is a testable hypothesis worthy of further inquiry. Coral cover on the lagoonal reefs in Bermuda and on the fore-reefs in the Florida Keys (which were all offshore reefs) did not exceed 30%, and therefore the level of interaction between coral colonies may have been fairly low. Percolation theory predicts that randomly distributed, equally-shaped objects will contact a large network of neighbors as the overall percent cover of the area approaches 60% (With et al. 1997). This implies that corals in habitats with greater than 60% coral cover should have at least one contacting neighbor (Figure 5.05), and probably more. Fore-reef sites at 15-m depth in Bermuda generally have over 75% coral cover (Murdoch unpublished data), and although the experiments have yet to be done, it may be that competitive exclusion structures coral assemblages in these high-density coral habitats.

283

Figure 5.05. A network of interacting corals on the Bermuda fore reef.

A functional-group approach also provided insight into the manner in which species were either dominant, subordinate or rare (e.g. Grime et al. 2001) across sites. The fact that species from one functional group dominate reefs under particular environmental conditions is contrary to the predictions of the unified neutral theory of Hubbell (2001), which predicts that all corals are equally likely to dominate any patch. The unified neutral theory was formulated for plant species within single functional groups and located within uniform habitats, such as species of hardwood trees within a flat area of tropical forest, and not for all of the plants that could coexist in the same forest. A key assumption of this dissertation is that corals as a group are like plants as a group, and furthermore that the functional groups of corals can be thought of as different “kinds” of corals the same way we think of trees, shrubs and weeds as different kinds of plants. If branched oviparous corals are functionally different from branched viviparous corals or any other

284 coral group, then it is inappropriate to assume that all corals will act in the same manner under the same conditions, just as we would not expect species of trees and grasses to share functional responses to the same environmental conditions. As originally proposed by Grime (1977) the Adaptive Strategies Theory focuses attention to the corners of the a triangular state-space model where each point of the triangle represents an extreme habitat and corresponding adapted functional group. IN this triangular model, intermediate habitats are predicted to be occupied by intermediate functional groups. If this is the case then functional groups will replace each other from habitat type to habitat type and the functional diversity across habitats will either be one (1) or none (0). Contrary to prediction, functional groups of corals showed a nested pattern of distribution, in which more than one functional group tended to exist within a habitat. It seems, based on this information, that the C, S and R strategies are complementary, not exclusionary. Species within the “Competitive Dominant” functional group may compete heavily with each other, but the effect of competition between functional groups appears to be reduced by trade-offs in the relative degree to which each functional group displays the three primary traits of growth, reproduction and defense. If so, the “Competitive Dominant” functional group appears to be badly named, and should instead be called the “Growth” functional group. A bigger problem with the Adaptive Strategies Theory is in defining the degree of resource availability and disturbance that a habitat possesses without tautologically referencing the biota about which one is attempting to make predictions. Also, the assumption that all sources of disturbance or resources can be delimited to one axis is probably false. In Bermuda each kind of disturbance affected each functional group

285 differently: the kinds of corals that can tolerate wave damage are different from those that can tolerate high levels of suspended sediment. Despite this problem, however, the Adaptive Strategy Theory does accurately predict that a ruderal strategy is required to persist under both kinds of disturbance, and also that the functional group with a competitive-dominant strategy would not be able to persist under either form of disturbance. The Adaptive Strategy Theory is typically used to predict the assemblage structure of only one upper-level taxon of organisms, such as plants, macroalgae or reef corals, ignoring all the other sessile organisms. It seems likely that species from multiple phyla, or even different kingdoms (i.e. sponges, soft corals, algae, hard corals etc), share membership in the same environmental functional group. Organisms adapted to the same habitat type should share functional traits. These larger taxonomic groupings may have had a long evolutionary history of interphyletic interaction, and, therefore such should be considered together. In the end, the weakness of the Adaptive Strategies Theory is that in simplifying complex ecological data, it also reduces the accuracy of the resulting information. This is of course also its strength. The Adaptive Strategies Theory provides a series of simple, testable hypotheses that can be used to guide ecological research in an iterative and informative manner. To fully test all of the predictions of Adaptive Strategies Theory, one would have to measure a very large number of traits in many species, as well as a broad range of physical, chemical, geological and biological characteristics across many habitats. The Adaptive Strategies Theory is not so much an accurate model of reality as it

286 is a powerful theoretical framework, which can be modified to give it great heuristic value for guiding ecological research.

287

288

REFERENCES

289

REFERENCES

Acevedo R, Morelock J, Olivieri RA (1989) Modification of coral reef zonation by terrigenous sediment stress. Palaios 4:92-100. Agassiz A (1895) A visit to the Bermudas in March 1894. Bulletin of the Museum of Comparative Zoology 26:209-281. Airoldi L (1998) Roles of disturbance, sediment stress, and substratum retention on spatial dominance in algal turf. Ecology 79:2759–2770. Ackleson (2003) Light in shallow water: A brief research review. Limnology and Oceanography 48:323-328. Andersen AN (1995) A classification of Australian ant communities, based on functional groups which parallel plant life-forms in relation to stress and disturbance. Journal of Biogeography 22:15-29. Anthony KRN (1999) Coral suspension feeding on fine particulate matter. Journal of Experimental Marine Biology and Ecology 232:85-106. Aronson RB, Edmunds PJ, Precht WF, Swanson DW, Levitan DR, (1994) Large-scale, long-term monitoring of Caribbean coral reefs: simple, quick. inexpensive techniques. Atoll Research Bulletin, 421: 145-155.

290 Aronson RB, Precht WF (1995) Landscape patterns of coral reef diversity: A test of the intermediate disturbance hypothesis. Journal of Experimental Marine Biology and Ecology 192:1-14. Aronson RB, Precht WF (1997) Stasis, biological disturbance, and community structure of a Holocene coral reef. Paleobiology 23:326-346. Aronson, RB, Swanson DW 1997. Video surveys of coral reefs: Uni- and multivariate applications. Proceedings of the 8th International Coral Reef Symposium, 14411446. Aronson RB, Precht WF (2001) White-band disease and the changing face of Caribbean coral reefs. Hydrobiologia 460:25-38. Aronson RB, MacIntyre IG, Precht WF, Murdoch TJT (2002) The expanding scale of species turnover events on coral reefs in Belize. Ecological Monographs 72:233250. Aronson RB, Precht WF (2006) Conservation, precaution, and Caribbean reefs. Coral Reefs 25:441-450. Bak RPM Engel MS (1979) Distribution, abundance and survival of juvenile hermatypic corals (Scleractinia) and the importance of life history strategies in the parent coral community. Marine Biology 54: 341-352. Baums IB, Miller MW, Szmant AM (2003) Ecology of a corallivorous gastropod, Coralliophila abbreviata, on two scleractinian hosts. I. Population structure of snails and corals. Marine Biology 142:1083-1091. Barnes D J (1973) Growth in colonial scleractinians. Bulletin of Marine Science 23:28098.

291 Bazzazz FA, Chiariello NR, Coley PD Pitelka LF (1987) Allocating resources to reproduction and defense. Bioscience 37:58–67. Bellwood DR, Hughes TP, Folke C, Nystrom M (2004) Confronting the coral reef crisis. Nature 6994:827-833. Beyer HL (2004) Hawth's Analysis Tools for ArcGIS: http://www.spatialecology.com/htools. 10/15/2007 Bolker BM, Pacala SW (1999) Spatial moment equations for plant competition: Understanding spatial strategies and the advantages of short dispersal. The American Naturalist 153:575-602. Bradbury RH, Loya Y (1978) A heuristic analysis of spatial patterns of hermatypic corals at Eilat, Red Sea. Am Nat 112:493-507. Brazeau DA, Gleason DF, Morgan ME (1998) Self-fertilization in brooding hermaphroditic Caribbean corals: evidence from molecular studies. Journal of Experimental Marine Biology and Ecology 231:225-238. Brooke S, Young CM (2003) Reproductive ecology of a deep-water scleractinian coral, Oculina varicosa, from the southeast Florida shelf. Continental Shelf Research 23:847-858. Browne B (1997) Coral bleaching: causes and consequences. Coral Reefs 16 Supplement: S129-S138. Bruno JF, Edmunds PJ (1997) Clonal variation for phenotypic plasticity in the coral Madracis mirabilis. Ecology 78:2177-2190. Budd AF, Foster Jr. CT, Dawson JP, Johnson KG (2001) The Neogene Marine Biota of Tropical America (“NMITA”) database: accounting for biodiversity in

292 paleoecology. Journal of Paleonology 75:743-751. Budd AF, Johnson KG, Stemmann T (2006) Neogene Marine Biota of Tropical America (NMITA) database of zooxanthellate corals. http://eusmilia.geology.uiowa.edu/database/corals/coralmnu.htm) Butler JN, Burnett-Herkes J, Barnes JA, Ward J (1993) The Bermuda Fisheries: A Tragedy of the Commons Averted? Environment 35/1/1, 1993: 6-33. CARICOMP (1997) Studies in Caribbean bleaching. In: Proceedings of the 8th International Coral Reef Symposium. 673-678. Carlon, D. B. and R. R. Olson. 1993. Short distance dispersal as an explanation of spatial pattern in two Caribbean reef corals. Journal Of Experimental Marine Biology and Ecology 173(2): 247-263. Carlon DB (1999) The evolution of mating systems in tropical reef corals. Trends in Ecology and Evolution 14:491-495. Chapin III, FS (1991) Integrated responses of plants to stress. BioScience 41:29-36. Chase JM, Leibold MA (2003). Ecological Niches: Linking Classical and Contemporary Approaches. University of Chicago Press, Chicago, IL. Chornesky EA (1991) The ties that bind: inter-clonal cooperation may help a fragile coral dominate shallow high-energy reefs. Marine Biology 109:41-51. Clarke KR, Gorley RN (2006) PRIMER v6: User manual/tutorial, PRIMER-E, Plymouth UK. Connell JH (1978) Diversity in tropical rain forests and coral reefs. Science 199:13021310.

293 Connell JH, Keough. MJ (1985) Disturbance and patch dynamics of subtidal marine animals on hard substrata. In S.T.A. Pickett and P.S. White (eds.), The Ecology of Natural Disturbance and Patch Dynamics, Academic Press, New York, pp.125151. Cook CB, Logan A, Ward J, Luckhurst B, Berg CJ (1990) Elevated temperatures and bleaching on a high-latitude coral reefL: the 1988 Bermuda event. Coral Reefs 9:45-49. De Mayer K (1998) Bonaire, Netherlands Antilles. In: Kjerfve B. (ed) UNESCO, 1998. CARICOMP – Caribbean coral reef, seagrass and mangrove sites.Coastal region and small island papers 3, UNESCO, Paris, xiv + 347 pp Dial R, Roughgarden J (1998) Theory of marine communities: the Intermediate Disturbance Hypothesis. Ecology 79:1412-1424. Dodge RE, Logan A, Antonius A (1982) Quantitative reef assessment studies in Bermuda: A comparison of methods and preliminary results. Bulletin of Marine Science, 32:745-760. Dodge RE, Vaisnys JR (1977) Coral populations and growth patterns: Responses to sedimentation and turbidity associated with dredging. Journal of Marine Research, 35:715-730. Dollar SJ (1982) Wave stress and coral community structure in Hawaii. Coral Reefs 1:7181. Done TJ (1982) Patterns in the distribution of coral communities across the cenbtral Great Barrier Reef. Coral Reefs 1:95-107.

294 Done TJ (1983) Coral zonation: its nature and significance. IN DJ Barnes (ed) Perspectives on Coral Reefs. Australian Institute of Marine Science 95-147. Dustan P, Halas JC (1987) Changes in the reef-coral community of Carysfort Reef, Key Large, Florida: 1974 to 1982. Coral Reefs 6:91-106. Edinger EN, Risk MJ (1995) Preferential survivorship of brooding corals in a regional extinction. Paleobiology 21:200-219. Edinger EN, Risk MJ (1999) Reef classification by coral morphology predicts coral reef conservation value.. Biological Conservation, 92, 1-13. Edmunds PJ, Bruno JF (1996) The importance of sampling scale in ecology: kilometerwide variation in coral reef communities. Marine Ecology Progress Series 143:165-171. Fadlallah YH (1983) Sexual reproduction, development and larval biology in scleractinian corals. Coral Reefs 2: 129-150. Fagerstromm JA (1988) A structural model for Reef Communities. Palaios 3:217-220. Falkowski PG, Jokiel PL, Kinzie III RA. (1990) Irradiance and Corals. In: Coral Reefs: Ecosystems of the World Vol 25 (Z. Dubinsky, ed.). Elsevier Scientific Publishing Co. Inc. New York:89-108. Fitzharding R (1985) Spatial and temporal variability in coral recruitment in Kaneohe Bay (Oahu, Hawaii). Proceedings of the 5th International Coral Reef Congress 4: 373-378. Flood VS, Pitt JM, Smith SR (2005) Historical and ecological analysis of coral communities in Castle Harbour (Bermuda) after more than a century of environmental perturbation. Marine Pollution Bulletin 5:545–557.

295 Fox BJ (1999) The genesis and development of guild assembly rules. Pages 23-57. In: E. Weiher and P. A. Keddy, eds. The search for assembly rules in ecological communities. Cambridge University Press, Cambridge. Fraser RH, Currie DJ (1996) The species richness-energy hypothesis in a system where historical factors are thought to prevail: coral reefs. American Naturalist 148:138159. Fricke H., Meischner D (1985) Depth limits of Bermudian scleractinian corals. Marine Biology, 88:175-187. Furla P, Allemand D, Shick JM, Ferrier-Pages C, Richier S, Plantivaux A, Merle P-L, Tambutte S (2005) The symbiotic anthozoan: A physiological chimera between alga and plant. Integrative and Comparative Biology 45:595-604. Gardner TA, Côte IM, Gill JA, Grant A, Watkinson AR (2003) Long-term region-wide declines in Caribbean Corals. Science 301:958-960. Garret P, Smith DK, Wilson OA, Patriquin D (1971) Physiography, ecology and sediments of two Bermuda patch reefs. Journal of Geology, 79:647-668. Garret P, Scoffin TP (1977) Sedimentation on Bermuda’s atoll rim. In: D. L. Taylor (ed) Proceedings of the Third International Coral Reef Symposium. 2:87-95. Rosenstiel School of Marine and Atmospheric Sciences, University of Miami, Miami FL, USA, 628 pp. Gause GF (1934) The struggle for existence. Williams and Wilkins, Baltimore Geister J. (1977) The influence of wave exposure on the ecological zonation of Caribbean coral reefs In: D.L. Taylor (ed.) Proceedings of Third International Coral Reef

296 Symposium Vol. 2: 23-29. Rosenstiel School of Marine and Atmospheric Science, Miami, Florida. Ginsburg RN, Shinn EA (1994) Preferential distribution of reefs in the Florida Reef Tract: the past is the key to the present. In:Ginsburg RN (compiler) Proceedings of the colloquium on global aspects of coral reefs: health, hazards and history, 1993. Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, pp 21-26. Gitay H, Noble IR. (1997). What are functional types and how should we seek them? In: T. Smith, H. Shugart and F. Woodward. (eds) Plant functional types: their relevance to ecosystem properties and global change.Cambridge, UK, Cambridge University Press: 3-19. Glynn PW (1988) Coral bleaching and mortality in the tropical eastern Pacific during the 1982-83 El Nino warming event. In: Ogden J, Wicklund R (eds) Mass bleaching of coral reefs in the Caribbean: a research strategy. NOAA’s Undersea Res Prog, St Croix, US Virgin Islands, Res Rpt 88-2:42-45 Glynn PW, Stewart RH (1973) Distribution of coral reefs in Pearl Islands (Gulf of Panama) in relation to thermal conditions. Limnology and Oceanography 18:367379. Goreau TF (1959) The ecology of Jamaican coral reefs: species composition and zonation. Ecology, Vol. 40, No. 1, pp. 67-90 Graus RR, Macintyre IG (1989) The zonation of Caibbean coral reefs as controlled by wave and light energy input, bathymetric setting and reef morphology: computer simulation experiments. Coral Reefs 8:8-18.

297 Green DG, Bradbury RH, Reichelt RE, (1987) Patterns of predictability in coral reef community structure. Coral Reefs, 6: 27-34. Grigg RW (1995) Coral reefs in an urban embayment in Hawaii: a complex case history controlled by both natural and anthropogenic stress. Coral Reefs, 14 253-266. Grime JP (1977) Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary theory. The American Naturalist 111: 1169-1194. Grime JP, Thompson K, Hunt R, Hodgson JG, Cornelissen JHC and thirty others (1997) integrated screening validates the primary axes of specialization in plants. Oikos 79:259-281. Grime J.P. 2001 Plant strategies, vegetation processes and ecosystem properties. John Wiley, Chichester. Gronbeck C (2002) Sol Path online software for sun aspect. http://susdesign.com/solpath/index.php Grottoli-Everett AG, Wellington GM (1997) Fish predation on the scleractinian coral Madracis mirabilis conrols its depth distribution in the Florida Keys, USA. Marine Ecology Progress Series 160:291-293. Hall VR, Hughes TP (1996) Reproductive strategies of modular organisms: Comparative studies of reef-building corals. Ecology 77:950-963. Harrison et al 1984 coral repro BS Harvell CD, Kim K, Burkholder JM, Colwell RR, Epstein PR, Grimes DJ, Hofman EE, Lipp EK, Osterhaus ADME, Overstreet RM, Porter JW, Smith GW, Vasta GR

298 (1999) Emerging marine diseases – Climate links and anthropogenic factors. Science 285:1505-1510. Heilprin A. (1889) The Bermuda Islands. Academy of Natural Sciences of Philadelphia, 231 pp. Published by the author. Heino and Kaitala 1999 resource use corals growth repro etc Highsmith RC (1979) Coral growth rates and environmental control of density banding. Journal of Experimental Marine Biology and Ecology 37: 105-125. Highsmith RC (1987) Reproduction by fragmentation in corals. Marine Ecology Progress Series, 7: 207-226. Hooper DU, Solan M, Symstad A, Diaz S, Gessner MO, Buchmann N, Degrange V, Grime P, Hulot F, Mermillod-Blondin F, Roy J, Spehn E, van Peer L (2002) Species diversity, functional diversity and ecosystem functioning. Chapter 17 in Loreau M, Naeem S, Inchausti P (eds) Biodiversity and Ecosystem Functioning: Synthesis and Perspectives. Oxford University Press, Oxford. pp:195-208. Hubbard JAE, Pocock YP (1972) Sediment-rejection by recent scleractinian corals: a key to palaeo-environmental reconstruction. Geol Rundschau 61:598-626. Hubbell SP (1997) A unified theory of biogeography and relative species abundance and its application to tropical rain forests and coral reefs. Coral Reefs 16: S9-S21. Hubbell SP (2001) The Unified Neutral Theory of Biodiversity and Biogeography. Princeton University Press, New Jersey. 448 pp. Hubbell SP (2005) Neutral theory in ecology and the hypothesis of functional equivalence Functional Ecology 19: 166-177.

299 Hudson JH (1985) Long-term growth rates of Porites lutea before and after nuclear testing: Enewetak Atoll (Marshall Islands). Proceedings of the 5th International Coral Reef Symposium 6: 179-185. Hughes TP (1984) Population dynamics based on individual size rather than age: a general model with a reef coral example. American Naturalist 123:778-795. Hughes TP, Baird AH, Dinsdale EA, Moltschaniwskij NA, Pratchett MS, Tanner JE, Willis BL (1999) Patterns of recruitment and abundance of corals along the Great Barrier Reef. Nature 397:59-63. Hughes TP, Baird AH, Bellwood DR, Card M, Connolly SR, Folke C, Grosberg R, Hoegh-Guldberg O, Jackson JBC, Kleypas J, Lough JM, Marshall P, Nyström M, Palumbi SR, Pandolfi JM, Rosen B, Roughgarden J (2003) Climate change, human impacts, and the resilience of coral reefs. Science 301(5635): 929-934. Hughes TP, Bellwood DR, Folke C, Steneck RS, Wilson JE (2005) New paradigms for supporting resilience of marine ecosystems. Trends in Ecology and Evolution. 20:380-386. Hughes, T. P. & J. H. Connell. 1999. Multiple stressors on coral reefs: a long term perspective. Limnology and Oceanography 44:932-940 Huston MA (1979) A general hypothesis of species diversity. American Naturalist 113:81-101. Huston MA (1985a) Patterns of species diversity on coral reefs. Annual Review of Ecology and Systematics 16:149-177. Huston MA (1985b) Variation in coral growth rates with depth at Discovery Bay, Jamaica. Coral Reefs 4:19-25.

300 Huston MA (1994) Biological Diversity: The coexistence of species on changing landscapes. Cambridge University Press, Cambridge UK. pp. 679. Huston MA, Smith T (1987) Plant succession, life history and competition. American Naturalist 130: 168-198. Jackson JBC (1979) Morphological strategies of sessile animals. In: Larwood G, Rosen BR (eds) Biology and systematics of colonial organisms. Academic Press, New York, pp 499-555. Jackson JBC (1991) Adaptation and diversity of reef corals. BioScience 41:475-482. Jackson JBC, Kirby MX, Berger WH, Bjorndal KA, Botsford LW, Bourque BJ, R.Bradbury RH, Cooke R, Erlandson J, Estes JA, Hughes TP, Kidwell S, Lange CB, Lenihan HS, Pandolfi JM, Peterson CH, Steneck RS, Tegner MJ, Warner RR Historical Overfishing and the Recent Collapse of Coastal Ecosystems Science 293: 629-637. Johannes RE, Wiebe WJ, Crossland CJ, Rimmer DW, Smith SV (1983) Latitudinal limits to coral reef growth. Marine Ecology Progress Series 11:105-111. Johnson KG, Budd AF, Stemann TA (1995) Extinction selectivity and ecology of Neogene Caribbean reef corals. Paleobiology 23(1): 52-73. Jones CG, Lawton JH, Shachak M (1994) Organisms as ecosystem engineers. Oikos 69: 373-386. Bermuda Institute for Ocean Sciences. Jones R, Pitt J, de Putron S, Venn A, Knap A, Noyes T, Evered J (2007) Marine Environmental Program Annual Report: 2005-2006. Kautsky L, Kautsky H (1989) Algal species diversity and dominance along gradients of stress and disturbance in marine environments. Plant Ecology 83:259-267.

301 Keddy PA (1992) Assembly and response rules: two goals for predictive community ecology. Journal of Vegetation Science 3: 157-164. Keough 1984 species area curves coral reefs Kirk, J. T. O., Light and Photosynthesis in Aquatic Ecosystems, Cambridge University Press, 1994. Knowlton N (2001) The future of coral reefs. Proceedings of the National Academy of Science 98(10): 5419-5425. Lang J (1973) Interspecific aggression by scleractinian corals. II. Why the race is not only to the swift. Bulletin of Marine Science, 23:260-279. Lavorel S, McIntyre S, Landsberg J, Forbes TDA (1997) Plant functional classification: from general groups to specific groups based on response to disturbance. Trends in Ecology and Evolution 12:474-478. Lewis JB, Snelgrove, P.V.R., 1990. Corallum morphology and composition of crustacean cryptofauna of the hermatypic coral Madracis mirabilis. Mar. Biol. 106, pp. 267– 272 Liddell WD, Ohlhorst SL (1987) Patterns of reef community structure, North Jamaica. Bulletin of Marine Science 40:311-329. Lidz BH, Hine AC, Shinn EA, Kindinger JL (1991a) Multiple outerreef tracts along the south Florida bank margin: outlier reefs, a new windward model. Geology 19:115-118. Lidz BH and Shinn EA (1991b) Paleoshorelines, reefs and a rising sea: South Florida, U.S.A. Journal of Coastal Research, 7(1): 203-229.

302 Lidz BH, Shinn EA, Hine AC, Locker SD (1997) Contrasts within an outlier-reef system: evidence for differential Quarternary evolution, south Florida windward margin, U.S.A. Journal of Coastal Research 13:711-731. Lirman D (2000) Fragmentation in the branching coral Acropora palmata (Lamark): growth, survivorship, and reproduction of colonies and fragments. Journal of Marine Biology and Ecology 251:41-57. Loehle C (1988) Problems with the triangular model for representing plant strategies. Ecology 69: 284-286. Logan A (1984) Interspecific aggression in hermatypic corals from Bermuda. Coral Reefs 3:131-138. Logan A (1988) Holocene reefs of Bermuda. Sedimenta IX, University of Miami, pp. 63. Logan A, Yang L, Tomascik T (1994) Linear extension rates in two species of Diploria from high latitude reefs in Bermuda. Coral Reefs 13:225-242. McCarthy JC, Pyle TE, Griffin GM (1974) Light transmissivity, suspended sediments and the legal definition of turbidity. Estuarine and Coastal Marine Science 2:291-299. McClanahan TR (2002) The near future of coral reefs. Environmental Conservation 29:460-483. McGill BJ, Enquist B, Weiher E, Westoby M (2006) Rebuilding community ecology from functional traits. Trends in Ecology and Evolution 21:178-185. Mayfield M, Daily G, Ackerly DD (2006) The functional diversity and conservation of herbaceous and shrubby plant communities in human-dominated landscapes. Journal of Ecology 94: 522-536.

303 Miller MW, Hay ME (1998) Effects of fish predation and seaweed competition on the survival and growth of corals. Oecologia 113:231-238. Mills MM, Lipschultz F, Sebens KP (2004) Particulate matter ingestion and associated nitrogen uptake by four species of scleractinian corals. Coral Reefs 23:311-323. Morris B, Barnes J, Brown F, Markham J (1977) The Bermuda Marine Environment. Special Publication 15, Bermuda Biological Station for Research, 120 pp Morse DE, Hooker N, Morse ANC, Jensen RA (1988) Control of larval metamorphosis and recruitment in sympatric agariciid corals. Journal of Experimental Marine Biology and Ecology 116:193–217 Murdoch TJT, Aronson (1999) Scale-dependent spatial variability of coral assemblages along the Florida Reef Tract. Coral Reefs 18:341-351. Nugues MM, Roberts CM (2003) Partial mortality in massive reef corals as an indicator of sediment stress on coral reefs. Marine Pollution Bulletin 46:314-323. Nystrom M, Folke C (2001) Spatial resilience of coral reefs, Ecosystems 4:406-417. O'Neill RV (1989) Perspectives in hierarchy and scale. Pages 140-156 in J. Roughgarden, R. M. May and S. A. Levin, eds. Perspectives in ecological theory. Princeton University Press, Princeton, NJ. . O’Neill RV, Milne BT, Turner MG, Gardner RH (1988) Resource utilization scales and landscape pattern. Landscape Ecology 2:63-69. Parker C., Oxenford HA (1998) Barbados. CARICOMP - Caribbean Coral Reef, Seagrass And Mangrove Sites. Edited By B. Kjerve. UNESCO, Paris. Peterson G, Allen CR, Holling CS (1998) Ecological Resilience, Biodiversity, and Scale. Ecosystems 1:6-18.

304 Porter JW (1976) Autotrophy, heterotrophy and resource partitioning in Caribbean reef building corals. American Naturalist, 110: 731-742. Porter JW, Meier OW (1992) Quantification of loss and change in Floridian reef coral populations. American Zoologist 32:625-640. Precht WF (1994) The use of the term guild in coral reef ecology and paleoecology: a critical evaluation. Coral Reefs 13:135-136. Precht WF, Aronson RB (2004) Climate flickers and range shifts of reef corals. Frontiers in Ecology and the Environment 2:307-314. Richmond RH (1997) Reproduction and recruitment in corals: Critical links in the persistence of reefs. In: C. Birkeland (ed). Life and death of coral reefs. Chapman & Hall, New York:175-197 Richmond RH, Hunter CL (1990) Reproduction and recruitment of corals: Comparisons among the Caribbean, the tropical Pacific, and the Red Sea. Marine Ecology Progress Series 60:185-203. Rogers CS (1979) The effect of shading on coral reef structure and function. Journal of Experimental Marine Biology and Ecology 41: 269-288 Rogers CS (1990) Reponses of coral reefs and reef organisms to sedimentation. Marine Ecology Progress Series 62: 185-202. Rogers CS, Fitz HCI, Gilnack M, Beets J, Hardin J (1984) Scleractinian coral recruitment patterns at Salt River submarine canyon, St Croix, US Virgin Islands. Coral Reefs 3: 69-76. Rogers CS, Garrison G, Grober R, Hillis ZM, Franks MA (1994) Coral reef monitoring manual for the Caribbean and western Atlantic. Virgin Islands National Park, St.

305 John, U.S, Virgin Islands. 107 pp Rotjan RD, Lewis SM (2005) Selective predation by parrotfishes on the reef coral Porites astreoides. Marine Ecology Progress Series 305:193-201. Rylaarsdam KW (1980) Life histories and abundance patterns of some common Caribbean reef corals. PhD dissertation. John Hopkins University, Baltimore, MD. Rylaarsdam KW (1983) Life histories and abundance patterns of colonial corals on Jamaican reef. Marine Ecolology Progress Series 13: 249-260. Sammarco PW (1985) The Great Barrier Reef vs. the Caribbean: comparisons of grazers, coral recruitment patterns and reef recovery. Pagers 391-397 in C. Cabrie and M. Harmelin-Vivien (eds) Proceedings of the fifth international Coral reef congress, Tahiti. Volume 4. Antenne Museum-EPHE, Moorea, French Polynesia. Sherman CDH, Ayre DJ, Miller KJ (2006) Asexual reproduction does not produce clonal populations of the brooding coral Pocillopra damicornis on the Great Barrier Reef, Australia. Coral Reefs 25:7-18. Shinn EA, Lidz BH, Kindinger JL, Hudson JH, Halley RB (1989) Reefs of Florida and the Dry Tortugas: Field Trip Guidebook T176, 28th International Geological Congress. American Geophysical Union, Washington, DC, 53 pp. Smith NP (1994) Long-term Gulf-to-Atlantic transport through tidal channels in the Florida Keys. Bulletin of Marine Science 54:602-609. Smith SR (1992) Patterns of coral recruitment and post-settlement mortality on Bermuda’s reefs: comparison to Caribbean and Pacific reefs. Am Zool 32:663673.

306 Smith SR (1999) Bermuda. In: Kjerve, N.J. (Ed.), CARICOMP: Coral Reef, Seagrass and Mangrove Sites. UNESCO Coastal Region and Small Island Papers, 3:247–257. Smith TM, Shugart HH, Woodward FT, Burton PJ (1993) Plant functional types. In: A.M. Solomon & H.H. Shugart (eds) Vegetation Dynamics and Global Change, Chapman and Hall, New York: 272-292. Smith T, Shugart H, Woodward F [eds] (2002) Plant functional types: their relevance to ecosystem properties and global change.Cambridge, UK, Cambridge University Press: 369 pp. Smith-Vaniz WF, Collette BB, Luckhurst BE (1999) Fishes of Bermuda: History, zoogeography, annotated checklist and identification keys. American Society of Ichthyologists and Herpetologists Special publication No. 4. Island Press, Inc., Lawrence Kansas. Solbrig OT (1993) Plant traits and adaptive strategies: Their role in ecosystem function. In E.-D. Schulze and H. A. Mooney (eds.), Biodiversity and Ecosystem Function, pp. 97-116. Springer-Verlag, Berlin. Soong K (1991) Sexual reproductive patterns of shallow-water reef corals in Panama. Bulletin of Marine Science 49(3): 832-846. Soong K (1993) Size as a species characteristic in massive reef corals. Coral Reefs 12: 77-83. Soong K, Lang JC (1992) Reproductive integration in reef corals. Biological Bulletin 183:418-431. Sorokin YI (1993) Coral Reef Ecology. Springer-Verlag New York pp.461.

307 Sousa WP (1979) Disturbance in marine intertidal boulder fields: the nonequilibriam maintenance of species diversity. Ecology 60:1225-1239. Steneck RS, Dethier MN (1994) A functional group approach to the structure of algaldominated communities. Oikos, 69: 476-498 . Sterrer W (1998) How many species are there in Bermuda? Bulletin of Marine Science 62: 809-840. Stoddart JA (1983) Asexual production of planulae in the coral Pocillopora damicornis. Marine Biology 76:279-284. Stoddart DR, Scoffin TP (1979) Microatolls: review of form, origin and function. Atoll Research Bulletin 224:1-17. Sullivan KM, Chiappone M (1993) Hierarchical methods and sampling design for conservation monitoring of tropical marine hard bottom communities. Aquat Cons Mar Freshwat Ecosys 3:169-187. Szmant AM (1986) Reproductive ecology of Caribbean reef corals. Coral Reefs, 5: 4354. . Szmant AM, Forrester A (1996) Water column and sediment nitrogen and phosphorus distribution patterns in the Florida Keys, USA. Coral Reefs 15:21}41. Thomas MLH (1985) Littoral community structure and zonation on the rocky shores of Bermuda. Bulletin Marine Science 38:857–870. Thomas, M. L., A. Logan. 1992. A Guide to the Ecology of the Shoreline and ShallowWater Marine Communities of Bermuda. Special Publication 30, Bermuda Biological Station for Research 345pp.

308 Tilman D (1982) Resource competition and community structure. Monographs in Population Biology 17, Princeton University Press 296 pp. Tilman D (1988) Plant strategies and the dynamics and structure of plant communities. Monographs in Population Biology 26, Princeton University Press 376 pp. Tunnicliffe V (1981) Breakage and propagation of the stony coral Acropora cervicornis. Proceeding of the National Academy of Science 78:2427-2431. UNESCO, 1998. CARICOMP – Caribbean coral reef, seagrass and mangrove sites. Coastal region and small island papers 3, UNESCO, Paris, xiv + 347 pp. Upchurch SB (1970) Sedimentation on the Bermuda platform. Northwestern University, Evanston, Illinois.206 pp. Veron JEN, Stafford Smith M (2000) Corals of the World. Australian Institute of Marine Sciences 1350 pp. Verrill AE (1902) The Bermuda Islands: Their scenery, climate, productions, physiography, natural history and geology, with sketches of their early history and the changes due to man. Trans. Conn. Acad. Arts Sci. 11:413-956. von Bodungen, B, Jickells TD, Smith SR, Ward JAD, Hillier GB (1982) The Bermuda Marine Environment, Vol. III. Special Publication 18, Bermuda Biological Station for Research, 123 pp. Walker ND, Roberts HH, Rouse LJ, Huh OK (1982) Thermal history of reef-associated environments during a record cold-air outbreak event. Coral Reefs 1:83-87. Walker, B. H., A. Kinzig, and J. Langridge. 1999. Plant attribute diversity, resilience, and ecosystem function: the nature and significance of dominant and minor species. Ecosystems 2:95-113.

309 Wallace CC (1985) Reproduction, recruitment and fragmentation in nin sympatric species of the coral genus Acropora. Marine Biology 88: 217-233. Wang J, van de Kreeke J, Krishnan N, Smith D 1994. Wind and tide response in Florida Bay. Bulletin of Marine Science 54: 579-601. Weber JN,White EW, Weber PH (1975) Correlation of density banding i n reef coral skeletons with environmental parameters: The basis for interpretation of chronological records preserved i n the coralla of corals. Journal of Paleobiology 1:137-149. Weiher, E, Clarke GDP, Keddy PA (1998) Community assembly rules, morphological dispersion, and the coexistence of plant species. Oikos 81: 309-321. Whittaker RH (1967) Gradient analysis in vegetation. Biological Review 42:207-264. Williams GC (1975) Sex and evolution. Princeton University Press, Princeton, NJ. 200pp. Wilkinson c (2004) Status of coral reefs of the world: 2004. Australian Institute of Marine Sciences, Townesville, Queensland. Wilson JB (1989) A null model of guild proportionality, applied to stratification of a New Zealand temperate rain forest. Oecologia 80:263-267. Wilson SD, Keddy PA (1986) Species competitive ability and position along a natural stress/disturbance gradient. Ecology 67:1236 1242. Wilson JB Lee WG (2000) C-S-R Triangle theory: community-level predictions, tests, evaluation of criticisms, and relation to other theories. Oikos, 91, 77-96. Woodley JD and 19 others (1981) Hurricane Allen’s impact on Jamaican coral reefs. Science 214:749-755.

310

311

APPENDICES

312 APPENDIX A: Digital Video Image Capture Methodology

•Introduction The following methodologies allow image capture, image manipulation and data analysis transects of spur-and-groove reef habitats that were filmed with a digital video camera in an underwater housing. While the methods were developed using Macintosh computers and associated hardware and software, the same or similar products exist for the PC, and most of the methods in this document can be accomplished with either platform.

• Hardware Required Sony DCR-VX1000 Digital Video Camera Amphibico VX1000 U/W Housing Macintosh Power Mac G3/266 Desktop Computer Radius Digital Video Card With Firewire Port

• Software Required Microsoft Excel 98 Adobe Photoshop 5.0 Radius PhotoDV plugin for Photoshop Applescript Script Editor 1.1.2.

IMPORTANT: Macros such as Applescript are programs that automate applications including the “Finder” application. As such it is possible to create, alter or delete files stored on any hard drive or unlocked disk attached to the computer while running a Macro. The manner in which Applescript macros work, and the

313 function of the macrosprovided should be understood before running them. The authors assume no responsibility for any type of loss, or any other damages, including, but not limited to special, incidental, consequential, or other damages. • Make Folder of Random Dot Files (Excel and Photoshop)

Turn on computer. Make new folder named “NewDots” on the startup HD. This is to hold the 99 random-dot image files.

Open Excel File named: random dot maker98 Open Adobe Photoshop 5.0 Open Applescript program: Random-dot Image Generator Run Applescript program: Random-dot Image Generator

Once this program has run – there should be 99 image files of random dot images with 10 random dots – lettered – in the folder named “NewDots” on your startup HD.

• Capture DV reef frames

If the digital video camera or cassette recorder is not attached to the computer – then shut down the Mac and connect the digital video cassette player to the Radius digital video card that is inside the Mac Power PC G3 via the Firewire ports.

Turn on the Mac and the DV cassette player.

314

If the digital cassette recorder is attached to the computer – quit Applescript and open the applications Adobe Photoshop 5.0 with the Radius plugin, and Microsoft Excel

Locate the correct video tape and place it in the digital video cassette player.

In Photoshop, open the Radius digital video plugin via the following path File: Import: Radius PhotoDV…

In the Radius control window play the videotape and record the start and stop times for each transect in the Excel spreadsheet named: Tape Gap Calculator

Rewind the tape and cue it to the first frame of the first transect.

Select the following options (see below) while still in the Radius Plugin:

Capture Mode: Autocapture

Every [XXX} frames i.e. Number of frames per time-gap between frame grabs – as prescribed by the Tape Gap Calculator

315 Until [52] are captured (give or take a couple)

Capture as[Yr-Si-D-Tr-Qu} - 2 digit yr, 2-5 letter site designation, 1 letter depth designation, 2 digit transect, 2 digit frame – starting at “00” i.e. 98-Pel-D-04-00 = Year 1998, Pelican Reef, Deep Site, 4th Transect, 1st frame Capture size: [720x480 (raw)] De-interlace: [None] Format:: [NTSC 4:3]

Once this is all in order – press start and wait for the 53 frames to be captured.

316 • Clean up video frames Once the 53 frames have been captured – they can all be adjusted to the correct size, color adjusted and saved by running the Photoshop action “DO PING 52 TIMES” ***

*** NB – it is important to re-record the sub-action “SAVE” within this action so that the files are saved to the correct folder.

The entire process of capturing and cleaning up the frames can then be repeated for the other 9 transects on the 3 digital video tapes per site.

Be sure to save all of the files from each transect in a unique folder nested within another folder holding all of the transects of a site– so that all of the files can be batch processed in the next step. Folders should be named in the following manner –

HARD DRIVE: SITE FOLDER: containing: HARD DRIVE: SITE FOLDER: TRANSECT FOLDER

i.e. MyHardDrive: 98-Pel-D 01-10  containing  98 Pel D 01 : 98 Pel D 10

• Paste dots to images

317

Once all of the frames are captured, cleaned up and saved into folders corresponding to site and transect – run the Applescript “Paste Random Dots” to paste the dots from randomly selected images to each of the 500+ frames of each site. This program will lead you through the selection of the folder containing the random-dot files, as well through to the selection of the first image file in the first of the 10 folders within the folder corresponding to the site currently under analysis.

• Analyze Video Frames Data can be either entered directly into an Excel spreadsheet on the computer – or entered onto a paper spreadsheet. The user records the sessile biota or substrate underneath the center point of the hollow square dots located on each image. The user also needs to record the depth of each frame.

If a randomly positioned dot falls on the depth gauge present in the image – the user can rotate the layer containing the dots 180° - this usually moves the dots to a position over the substrate.

Several Photoshop actions are included that facilitate in the analysis of the video images. The quick-keys corresponding to these actions are:

F1

– Select Background Layer

F2

– Select Layer 1

318 F3

– Toggle Brightness/Contrast manipulation control

�-F4 – Flatten layers down to background F5

– Rotate currently selected layer 180°

F15

– “PING” – a suite of actions that manipulate the raw video image so it is

the correct size and of improved resolution and color balance. �-L

- Levels control – a very useful tool for color-adjusting underwater images

⇑ -L

- Auto Levels adjust

Once all the images have been analyzed all of the images in each folder nested in the folder for each site can be flattened using the Batch command and the “flatten” action in Photoshop: The menu commands are as follows:

File: Automate: Batch…

319

• Burn to CD (E) Once an entire reef site has been analyzed, the digital video frames and the Excel spreadsheets and other statistical data can be stored in a folder and subsequently copied onto a CD using CD recording software such as Adaptec Toast 3.5.3

320 APPENDIX B: Applescript computer program for using Microsoft Excel® and Adobe Photoshop® software to place dots on frames

-- the list of file types which will be processed -- eg: {"PICT", "JPEG", "TIFF", "GIF"} property type_list : {"8BPS", "PICT", "JPEG", "TIFF", "GIF"} -- since file types are optional in Mac OS X, -- check the name extension if there is no file type -- NOTE: do not use periods (.) with the items in the name extensions list -- eg: {"txt", "text", "jpg", "jpeg"}, NOT: {".txt", ".text", ".jpg", ".jpeg"} property extension_list : {} property timerRoutine : ""

-- This droplet processes both files or folders of files dropped onto the applet on open these_items

----the below checks to ensure needed Excel files and Photoshop Actions are in place tell application "Finder" activate set ButtonChoice to display dialog "

••••WARNING••••

321

This program will modify ALL Photoshop files contained within the files or folders dragged over its icon.

Do you wish to continue?" buttons ¬ {"STOP!", "OK"} default button "STOP!" with icon caution

set button_name to button returned of ButtonChoice if button_name is "STOP!" then return end if

set ButtonChoice to display dialog "This program requires that you have the included Microsoft Excel file random dot maker (25) open, and the included Photoshop Action file placed in the Photoshop Extensions Folder on your HD.

Have you done this?" buttons ¬ {"NO! Quit so I can do it", "OK"} default button "OK" with icon note end tell

322 set button_name to button returned of ButtonChoice if button_name is "NO! Quit so I can do it" then return end if

repeat with i from 1 to the count of these_items set this_item to (item i of these_items) set the item_info to info for this_item if folder of the item_info is true then

process_folder(this_item)

else if (alias of the item_info is false) and ¬ (the file type of the item_info is in the type_list) then

process_item(this_item)

else tell application "Finder" activate display dialog "The files or folders did not contain Photoshop files." buttons {"Quit"} ¬

323 default button "Quit" with icon stop end tell return end if end repeat

tell application "Finder" activate display dialog "The session has finished." buttons {"Quit"} ¬ default button "Quit" with icon stop end tell end open

-- this sub-routine processes folders on process_folder(this_folder) set these_items to list folder this_folder without invisibles repeat with i from 1 to the count of these_items set this_item to alias ((this_folder as text) & (item i of these_items)) set the item_info to info for this_item if folder of the item_info is true then process_folder(this_item) else if (alias of the item_info is false) and ¬ (the file type of the item_info is in the type_list) then -- or ¬

324 --the name extension of the item_info is in the extension_list) then process_item(this_item) end if end repeat end process_folder

-- this sub-routine processes files on process_item(this_item) -- NOTE that the variable this_item is a file reference in alias format -- FILE PROCESSING STATEMENTS GOES HERE

tell application "Microsoft Excel" Activate (*set Visible of ActiveWindow to false*) Activate Window "random dot maker FGB (25)" Select Range "R1C1" CopyObject Selection Paste Activate ChartObject "Chart 13" of ActiveSheet Select ChartArea of ActiveChart set CutCopyMode to false CopyObject ChartArea of ActiveChart end tell

325

(* this part is not needed tell application "Adobe® Photoshop® 6.0.1" activate do script "Dot Grabber" -- an Adobe Photoshop 6.0 action - see attached end tell *)

-- this selects the next photoshop file tell application "Finder" activate select file (this_item) open selection end tell

-- this pastes the NewDots onto a new layer on the video frame, saves the file and closes it tell application "Adobe® Photoshop® 6.0.1" activate do script "Dot Paster" end tell end process_item

326 Appendix C: A list of coral species observed in Bermuda The following is a list of the species of scleractinian hard corals that have been observed in Bermuda by the author. Just list the ones seen at your sites

Branched Oviparous 1. Oculina diffusa 2.

Oculina robusta ???

Branched Viviparous 1. Madracis decactis 2. Madracis formosa (New possible record: S. R. Smith) 3. Madracis mirabilis 4. Porites furcata (New record: T.J.T. Murdoch: unpublished confirmation) 5. Porites porites Massive Viviparous 1. Agaricia agaricites (T.J.T. Murdoch, A. Venn: possible sighting) 2. Dichocoenia stokesii 3. Favia fragum 4. Isophyllia sinuosa 5. Meandrina meandrites 6. Porites astreoides 7. Siderastrea radians Massive Oviparous 1. Diploria labyrinthiformis

327 2. Diploria strigosa 3. Montastraea cavernosa 4. Montastraea faveolata 5. Montastraea franksi 6. Montastraea species A (The majority of M. annularis spp. appear to be a hybrid between M. faveolata and M. franksi in BDA) 7. Siderastrea siderea (T.J.T. Murdoch, S. du Putron: possibly regionally extinct?) 8. Stephanocoenia intersepta Foliose, Plating and Solitary 1. Agaricia fragilis 2. Scolymia cubensis (W. Sterrer: potentially more than one species)

328 Appendix D: Bermuda climatology Climatology of Bermuda from 1949-1999, as published online by the Bermuda Weather Service (http://www.weather.bm/data/climatology.html)

329 Appendix E: Logistic regression of rank abundances; Florida data Logistic regression of rank abundances versus total abundance per transect for each functional group

Logistic Fit of BV By TCA 1.00

5 4

BV

0.75

3

0.50

0.25 2 0.00

1 ­10 0

10 20 30 40 50 60 70 80 90 100 110 120 130

SUM

Whole Model Test Model Difference Full Reduced

-LogLikelihood 6.34659 238.99612 245.34271

RSquare (U) Observations (or Sum Wgts) Converged by Objective

DF 1

0.0259 200

ChiSquare 12.69318

Prob>ChiSq 0.0004

330 Parameter Estimates Term Intercept Intercept Intercept Intercept TCA

Estimate -2.4993419 -0.1303332 2.07864736 4.94687655 -0.0150532

Std Error 0.3709632 0.2248103 0.2772554 0.6259174 0.0042367

ChiSquare 45.39 0.34 56.21 62.46 12.62

Prob>ChiSq <.0001 0.5621 <.0001 <.0001 0.0004

FP

Logistic Fit of FP By TCA 1.00

5

0.75

4

0.50 3 0.25

0.00

2 1 ­10 0

10 20 30 40 50 60 70 80 90 100 110 120 130

SUM

Whole Model Test Model Difference Full Reduced

-LogLikelihood 7.21570 217.89041 225.10611

RSquare (U) Observations (or Sum Wgts) Converged by Objective

DF 1

0.0321 200

ChiSquare 14.4314

Prob>ChiSq 0.0001

331 Parameter Estimates Term Intercept Intercept Intercept Intercept TCA

Estimate -3.3128101 -1.3570022 1.05681986 4.49707184 -0.0164569

Std Error 0.5286694 0.2650898 0.2432521 0.5152874 0.0043939

ChiSquare 39.27 26.20 18.88 76.17 14.03

Prob>ChiSq <.0001 <.0001 <.0001 <.0001 0.0002

Logistic Fit of MV By TCA 1.00

5 4 3

MV

0.75

0.50

2

0.25

0.00

1 ­10 0

10 20 30 40 50 60 70 80 90 100 110 120 130

SUM

Whole Model Test Model Difference Full Reduced

-LogLikelihood 10.41441 166.38296 176.79736

RSquare (U) Observations (or Sum Wgts) Converged by Gradient

DF 1

0.0589 200

ChiSquare 20.82881

Prob>ChiSq <.0001

332 Parameter Estimates Term Intercept Intercept Intercept Intercept TCA

Estimate -1.254137 2.70655809 4.91549566 5.85518535 -0.0227229

Std Error 0.2738426 0.3465647 0.5580352 0.7797963 0.0050557

ChiSquare 20.97 60.99 77.59 56.38 20.20

Prob>ChiSq <.0001 <.0001 <.0001 <.0001 <.0001

Logistic Fit of MO By TCA 1.00

4 3 2

MO

0.75

0.50

1

0.25

0.00

­10 0

10 20 30 40 50 60 70 80 90 100 110 120 130

SUM

Whole Model Test Model Difference Full Reduced

-LogLikelihood 21.677679 55.428898 77.106577

RSquare (U) Observations (or Sum Wgts) Converged by Objective

DF 1

0.2811 200

ChiSquare 43.35536

Prob>ChiSq <.0001

333 Parameter Estimates Term Intercept Intercept Intercept TCA

Estimate -0.251108 1.39009666 2.36232842 0.12926131

Std Error 0.4552401 0.5633292 0.7768345 0.0333187

ChiSquare 0.30 6.09 9.25 15.05

Prob>ChiSq 0.5812 0.0136 0.0024 0.0001

334

335

BIOGRAPHICAL SKETCH

Name of Author: Thaddeus James Thomas Murdoch Place of Birth:

Somerset Village, Sandy’s Parish, BERMUDA

Date of Birth:

April 18, 1966

Graduate and Undergraduate Schools Attended: University of South Alabama, Mobile, Alabama, USA Dalhousie University, Halifax, Nova Scotia, CANADA Degrees Awarded: 1995 – 1998

Master of Science in Marine Science, University of South Alabama, Mobile, Alabama, USA and the Dauphin Island Sea Lab, Dauphin Island, Alabama, USA.

1988 - 1991

Bachelor of Arts, Honors in Psychology (Neuroendocrinology), Dalhousie University, Halifax, N.S. CND

1984 - 1988

Bachelor of Science in Biology, Dalhousie University, Halifax, N.S. CND

Awards and Honors:

Bermuda Biological Station for Research, Inc. - Grant in Aid: 2003 PADI Foundation: 2000 Ph.D. Fellowship, Marine Science Dept., U. South Alabama: 1999 to 2002 Nelson Award for Outstanding Masters Student, U. South Alabama: 1998 Marine Science Dept. Student Assistantship, U. South Alabama: 1995 to 1998