xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001919226
007 cr mnu|||uuuuu
008 071214s2007 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001918
Higham, Christopher John.
Modeling larval connectivity among coral habitats, Acropora palmata populations, and marine protected areas in the Florida Keys National Marine Sanctuary
h [electronic resource] /
by Christopher John Higham.
[Tampa, Fla] :
b University of South Florida,
The Florida Keys National Marine Sanctuary (FKNMS) encompasses North America's only living coral barrier reef and the third longest barrier reef in the world, making it a unique national treasure of international notoriety (FKNMS, 2005). Recent evidence of environmental decline within the sanctuary has created a sense of urgency to understand and protect the valuable resources within. This thesis contributed to the understanding of habitat connectivity to aid managers and decision makers in the creation of additional Marine Protected Areas (MPAs) in the FKNMS to help prevent further environmental decline. This research specifically focused on modeling larval transport and larval connectivity among Acropora palmata (Lamarck, 1816) populations, coral habitats and MPAs in the upper and middle FKNMS.^ ^The transport of larvae in relation to ocean currents is a very limited area of research, and the analytic modeling results may serve as powerful guides to decisions about the relative importance of individual coral habitats and MPAs in the study area.Larval transport was modeled with ArcGIS and TauDEM using SoFLA-HYCOM simulated ocean currents during the A. palmata spawning season. This model allowed for the assessment of coral habitat and A. palmata population larval connectivity. The dependence of three distant A. palmata test populations on other upstream coral habitats and A. palmata populations significantly differed (Kruskal-Wallis test, P less than 0.0001). The clonally diverse Sand Island Reef A. palmata population's larval connectivity was significantly higher compared to other distant monoclonal populations (Mann-Whitney test, P less than 0.0001).^ ^Compared to the clonal structure of each test population determined by Baums, Miller, and Hellberg (2006), results indicated simulated larval connectivity may be a determinant of A. palmata population clonal diversity.By modeling MPA and coral habitat connectivity, this study also identified unprotected and distant coral habitat areas with the greatest downstream influence on MPAs; these may serve as potential coral larvae sources. It is recommended that establishing these areas as no-take MPAs would improve overall coral habitat and MPA network connectivity.
Thesis (M.A.)--University of South Florida, 2007.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 134 pages.
Adviser: Paul Zandbergen, Ph.D.
D-infinity flow routing.
t USF Electronic Theses and Dissertations.
Modeling Larval Connectivity among Coral Habitats, Acropora palmata Populations, and Marine Protected Areas in the Flor ida Keys National Marine Sanctuary by Christopher John Higham A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts Department of Geography College of Arts and Sciences University of South Florida Major Professor: Paul Zandbergen, Ph.D. Jayajit Chakraborty, Ph.D. Susan Bell, Ph.D. Data of Approval: April 10, 2007 Keywords: GIS, ArcGIS, TauDEM, D flow routing, ocean cu rrents, flow direction, contributing flow, upstream dependen ce, clonal diversity, SoFLA-HYCOM Copyright 2007, Christopher John Higham
i Table of Contents List of Tables iv List of Figures vi Abstract ix Chapter One: Introduction 1 Background 1 Goal 8 Objectives and Null Hypotheses 8 Objective One 8 Objective Two 8 Objective Three 8 Objective Four 9 Objective Five 9 Objective Six 9 Chapter Organization 9 Chapter Two: Literature Review 11 History of MPAs 11 National MPA Center 12 General Design of MPA Networks 12 Applying Larval Transport Patterns to MPA Design 15 Effectiveness of MPAs 16 Connectivity 17 Measures of Connectivity in Spatial Ecology 17 Landscape Connectivity 18 Larval Transport and Dispersa l in the Marine Environment 20 Connectivity in the Marine Environment 23 Florida Keys Nationa l Marine Sanctuary 27 Background 27 MPAs 29 Oceanography 30 Larval Transport and Recruitment 32 Stony Coral Populations 34 Elkhorn Coral ( Acropora palmata ) 35
ii Chapter Three: Methods and Materials 38 Study Area 38 Data 41 Shoreline and Boundaries 41 Benthic Habitats 42 Coral Habitats 42 Acropora palmata Populations 43 Acropora palmata Population Genetics 44 Modeled Ocean Currents 45 Ocean Current Data Post-P rocessing and Interpolation 47 Point Feature Class Creation 47 Cross-validation of Interpolated Grids 48 Flow Direction Grid Calculations 49 A GIS-based Model of Larval Transport 52 Connectivity Analyses 54 Analysis One: Larval Connectivity and A. palmata Clonal Diversity 56 Analysis Two: Larval Connectivity and Unprotected Larvae Sources 63 Summary of Assumptions 68 Chapter Four: Results 71 Analysis One: Larval Connectivity and A. palmata Clonal Diversity 71 Modeled Larval Transport 71 Modeled Larval Connectivity 74 Statistical Analyses 80 Analysis Two: Larval Connectivity and Unprotected Larvae Sources 87 Modeled Larval Transport 87 Modeled Larval Connectivity 89 Statistical Analyses 91 Unprotected Larvae Sources 98 Chapter Five: Discussion and Conclusions 101 Introduction 101 Larval Connectivity a nd Clonal Diversity of A. palmata populations 103 Critical Unprotected Coral Habita t Upstream of Existing MPAs 104 Summary of Contributions 106 Summary of Limitations and Assumptions 107 Usefulness of this Research 108 Recommendations for Future Research 110 Literature Cited 113 Appendices 121 Appendix A 122 Appendix B 126 Appendix C 130
iii Appendix D 131 Appendix E 132 Appendix F 133 Appendix G 134
iv List of Tables Table 1 Descriptions of MPAs within the Study Area 41 Table 2 Locations of all Validated A. palmata Populations (Miller et al., in preparation) 44 Table 3 Three A. palmata Test Populations and Their Clonal Diversity 45 Table 4 Larval Source and Sink Areas in Analyses One and Two 54 Table 5 Summary Statistics of Augus t Contributing Flow from All Coral Habitats to Each A. palmata Test Population 82 Table 6 Summary Statistics of August Contributing Flow Only from Other Validated A. palmata Populations to Each A. palmata Test Population 84 Table 7 Kruskal-Wallis Test and Subsequent Mann-Whitney All-Pairwise Comparison Test for Differences in Larval Connectivity among Each A. palmata Test Population and All Other Coral Habitats 85 Table 8 Kruskal-Wallis Test and Subsequent Mann-Whitney All-Pairwise Comparison Test for Differences in Larval Connectivity among Each A. palmata Test Population and Other Validated A. palmata Populations 86 Table 9 Comparison of Differences in Larval Connectivity (Kruskal-Wallis mean ranks) and Clonal Popula tion Structure among the Three A. palmata Test Populations 87 Table 10 Summary Statistics of A ugust Contributing Flow from Distant Coral Habitats to MPAs, by Coral Habitat Type 95 Table 11 Summary Statistics of A ugust Contributing Area from Distant Coral Habitats to MPAs, by Coral Habitat Type 95
v Table 12 Kruskal-Wallis Test for Diffe rences in Coral Larval Connectivity with MPAs among Coral Habitat Types 97 Table 13 Summary Statistics of Larv al Connectivity and Contributing Area (by habitat type) among Each Re gion containing High Total August Larval Connect ivity with MPAs 100 Table A1 Cross-Validation of Interpolated Grids 125 Table C1 Daily Contributing Flow Valu es from All Coral Habitats to Each A. palmata Test Population 130 Table E1 Daily Contributing Flow Values from Validated A. palmata Populations to Each A. palmata Test Population 132 Table F1 Daily Contributing Flow Valu es from Each Coral Habitat Type to MPAs. 133 Table G1 Mann-Whitney All-Pairwise Comparison Test for Differences in Coral Larval Connectivity with MPAs among Coral Habitat Types 134
vi List of Figures Figure 1 Potential Scenarios for th e Spatial Connectedness of Distant Populations 24 Figure 2 A Colony of A. palmata (Photo Courtesy of NOAA Center for Coastal Monitoring and AssessmentÂ’s Biogeography Team) 36 Figure 3 Flow Diagram of the Typical Acropora sp. Spawning Cycle (Photos Courtesy of www.undersea.com.au) 37 Figure 4 Regional Map of the Flor ida Keys National Marine Sanctuary (Courtesy of FKNMS) 39 Figure 5 The 800 km2 Study Area in Northeastern FKNMS 40 Figure 6 Geographic Extent of SoFLA-HYCOM (Figure from Kourafalou et al., 2005) 46 Figure 7 SoFLA-HYCOM Data Post-Processing 50 Figure 8 D Method for Determination of Cu rrent Flow Directions (Figure from Tarboton, 2005) 51 Figure 9 Upslope or Upstream Depende nce Function of Grid Target Cells y (Figure from Tarboton, 2005) 53 Figure 10 Modeling Overview (See Greater Detail in Appendix B) 55 Figure 11 Larval Trans port Model Ran for Each A. palmata Test Population and Each Daily Current Regime during August 57 Figure 12A Flow Diagram of Analysis One 60 Figure 12B Flow Diagram of Analysis One 61 Figure 13 Flow Diagram of Analysis Two 65
vii Figure 14 Total August Contributing Flow Grid for the Sand Island Reef A. palmata Test Population 72 Figure 15 Total August C ontributing Flow Grid for the Little Grecian Reef A. palmata Test Population 73 Figure 16 Total August Contributing Flow Grid for the Horseshoe Reef A. palmata Test Population 74 Figure 17 One Daily Averaged SoFLA-HYCOM Simulation of Ocean Current Direction in Relation to Coral Habitat and A. palmata Test Population Locations 75 Figure 18 Coral Habitats and Validated A. palmata Populations 77 Figure 19 Total August Larval Conn ectivity among Coral Habitats and the Sand Island Reef A. palmata Test Population 78 Figure 20 Total August Larval Conn ectivity among Coral Habitats and the Little Grecian Reef A. palmata Test Population 79 Figure 21 Total August Larval Conn ectivity among Coral Habitats and the Horseshoe Reef A. palmata Test Population 80 Figure 22 Box-Plots of Mean Daily Contributing Flows from All Coral Habitats within 25 km to Each A. palmata Test Population 82 Figure 23 Box-Plots of Mean Daily Contributing Flows Only from Other Validated A. palmata Populations within 25 km to Each A. palmata Test Population 84 Figure 24 Marine Protected Areas within the Study Area 88 Figure 25 Total August Contri buting Flow to all MPAs 89 Figure 26 Total August Larval C onnectivity among Coral Habitats and MPAs 91 Figure 27 Ten Coral Habitat Types within the Study Area 92 Figure 28 Box-Plots of Mean Daily C ontributing Flow from Unprotected and Distant Coral Habitats to MPAs, by Coral Habitat Type 94 Figure 29 Contributing Area (km2) and Contributing Flow to MPAs (Mean Flow Fraction) among Coral Habitat Types 96
viii Figure 30 Unprotected a nd Distant Coral Habitat Regions with High Total August Larval Connect ivity with MPAs 98 Figure A1 SoFLA-HYCOM Current Vector Point Features 123 Figure A2 SoFLA-HYCOM U Current Vect or Component Spline Interpolation and Point Features used for Cross-Validation 124 Figure B1 Larval Transport and Connec tivity Analytic Model: Analyses One and Two 126 Figure B2 Larval Transport and Connec tivity Analytic Model: Analyses One and Two (Continued) 127 Figure B3 Larval Transport and Connec tivity Analytic Model: Analyses One and Two (Continued) 128 Figure B4 Larval Transport and Connec tivity Analytic Model: Analyses One and Two (Continued) 129 Figure D1 Description of Box-Plots (fro m Analyse-It for Microsoft Excel Help Index) 131
ix Modeling Larval Connectivity among Coral Habitats, Acropora palmata Populations, and Marine Protected Areas in the Florid a Keys National Marine Sanctuary Christopher John Higham ABSTRACT The Florida Keys National Marine Sanctuary (FKNMS) encompasses North AmericaÂ’s only living coral barrier reef and the third longest barrier reef in the world, making it a unique national treasure of intern ational notoriety (FKNMS, 2005). Recent evidence of environmental decline within the sanctuary has created a sense of urgency to understand and protect the valuab le resources within. This thesis contributed to the understanding of habitat connectivity to aid ma nagers and decision makers in the creation of additional Marine Protected Areas (MPAs) in the FKNMS to help prevent further environmental decline. This research specifically focused on modeling larval transport and larval connectivity among Acropora palmata (Lamarck, 1816) populations, coral habitats and MPAs in the upper and middle FKNMS. The transport of larvae in relation to ocean currents is a very limited area of research, a nd the analytic modeling results may serve as powerful guides to decisions about the relativ e importance of individual coral habitats and MPAs in the study area.
x Larval transport was modeled with ArcGIS and TauDEM using SoFLA-HYCOM simulated ocean currents during the A. palmata spawning season. This model allowed for the assessment of coral habitat and A. palmata population larval connectivity. The dependence of three distant A. palmata test populations on other upstream coral habitats and A. palmata populations significantly differed (K ruskal-Wallis test, P < 0.0001). The clonally diverse Sand Island Reef A. palmata populationÂ’s larval connectivity was significantly higher compared to other dist ant monoclonal populations (Mann-Whitney test, P < 0.0001). Compared to the clonal st ructure of each test population determined by Baums, Miller, and Hellberg (2006), results in dicated simulated larval connectivity may be a determinant of A. palmata population clonal diversity. By modeling MPA and coral habitat c onnectivity, this study also identified unprotected and distant coral ha bitat areas with th e greatest downstream influence on MPAs; these may serve as potential coral la rvae sources. It is recommended that establishing these areas as no-take MPAs would improve overall coral habitat and MPA network connectivity.
1 Chapter One: Introduction Background Geography is about the Earth and its feat ures. It is not only knowing about EarthÂ’s features themselves, but understandi ng the interdependence and connectivity of these features (Bell, 2005). The Florida Keys are a unique region of the world, where humans are highly dependent upon the KeyÂ’s environmental well-being. Humans are highly dependent upon the coral reefs and ot her habitats of the Florida Keys, both economically and socially. Ecosystems of the Florida Keys are in great decline, and if humans do not intervene and attempt to understand and protect these ecosystems, humankind may lose them forever. This is why understanding connec tivity in the Florida Keys is so critical; it will help us in our efforts to pres erve the relationships among the regionÂ’s humans and marine habitats, two very interdependent and important features of the Earth. A Geographic Information System (GIS) based analytic approach to learn about the interdependence of marine habitats will take us one step closer to understanding how we can help manage and prot ect these environmental resources. Marine Protected Areas (MPAs) are eff ective management tools for protecting natural and cultural re sources. Moilanen and Nieminen (2002) review many examples of how connectivity is a fundamental concept wide ly utilized in spatial ecology and resource management. Jackson and Massey (2006) desc ribe the value of thinking geographically;
2 how taking into account proximity, distance, in teractions, interdependencies, and scale when designing MPAs could significantly impact their effectiveness. Modeling ecological links (i.e., connec tivity) between MPAs is difficult due to complex biophysical relationships present in the ocean realm, but innovative technologies and refined spatial modeling tools have opened a new door in to this field of study. It is an immense challenge to understand marine ecosystem patte rns over spatial and temporal scales that are directly relevant to conservation a nd ecosystem management (Palumbi, Gaines, Leslie, & Warner, 2003). The challenge lies in numerous known and unknown variables one must consider when modeling dynamic ecological relationships within the mari ne environment, such as connectivity. Empirical data on the spatial connectedness of ecosystems are scarce for the marine environment when compared to th e terrestrial environment (Palumbi et al., 2003). One reason for the limitation is that ma rine larval biology a nd behavior is very complex; there are numerous larval stages, some species have active and/or passive swimming stages and the duration of time spen t drifting and/or swimming in the water column greatly varies among species also. The larval stage and swimming or drifting behavior within the water column, in addition to the effects of ocean currents (e.g., mixing, retention, and dispersal) create dyna mic and variable ecol ogical relationships much more difficult to quantify and understa nd. Recently, spatial modeling tools have begun to secure a greater unders tanding of marine connectivit y, and these tools can play an essential role in MPA science. Connectivity in this thesis specifically refers to a functiona l relationship defined as a spatial and ecological li nk between areas via larval tr ansport and ocean currents.
3 There is a great need for refined spatial models of larval transport and ocean currents to assess connectivity of MPAs. For example, th e transport of larvae in relation to ocean currents is a very limited area of researc h, and improved models will serve as powerful guides to decisions about the relative importance of indivi dual populations and/or MPAs to overall MPA network connectivity. Protecting natural and cultural resources are integral to MPA management. Executive Order 13158 (Federal Register, 2000) defines a MPA as Â“any area of the marine environment that has been reserved by Federal, State, territorial, tribal, or local laws or regulations to provide lasting protection for part or all of the natural and cultural resources therein.Â” There are many types of MPAs, each with different definitions based primarily on the level of protection provided by the MPA. For example, a marine reserve is defined as an area closed to fishing and other extractive activitie s (Meester, Mehrotra, Ault, & Baker, 2004). For the purpose of this thesis, all MPAs will be analyzed regardless of type assigned to each of them. According to Salm, Cl ark, & Siirila, (2000), MPAs Â“have been used effectively both na tionally and internationally to conserve biodiversity, manage natural resources, prot ect endangered species, reduce user conflicts, provide educational and research opportunitie s, and enhance commercial and recreational activitiesÂ”. Spatial modeling of ecosystem patterns has advanced, but there is much room for refinement in order to better understand c onnectivity between MPAs. There are over 50 examples of how the use of MPAs as mana gement tools enhanced marine communities within their boundaries; however, very little is known whether MPAs have measurable effects beyond their boundaries (Halpern, 2003; Palumbi, 2003). Enhancing nearby
4 populations through the transport of eggs a nd larvae produced in a MPA is a compelling yet unresolved aspect of MPAs fo r fishers and fisheries managers ( Kendall Jr. & Picquelle, 2003). Until recently, limitations on data availability and spatial modeling tools were major obstacles to understanding marine ecosystem patterns over spatial and temporal scales. A better understanding of marine ecosyst em patterns over spatial and temporal scales that are directly relevant to cons ervation and ecosystem management is badly needed. Spatial models must be refine d to enhance our knowledge of ecological relationships, such as connec tivity. It is common knowledge that through shared species and oceanographic processes, many marine ecosystems are intimately linked. The connections between a MPA and its surroundi ng ecosystems are mediated by the ocean environment and the life histories of the species present (Palumbi et al., 2003). Population distribution and abunda nce of marine organisms w ith complex life cycles are governed by a large variety of physical, chem ical and biological pr ocesses that occur on local, regional and global scales (Thibaut, Lagadeuc, Olivier, Dauvin, & Retire, 1998). These natural variables alone add complexity to the challenge of assessing connectivity, but human action or inaction in one MPA can also have consequences for the shared living organisms occupying these areas with no definite boundaries (Morgan, Etnoyer, Wilkinson, Herrmann, Tsao, & Maxwell, 2003). Recent advances in technologies ar e helping improve upon MPA research, planning, and management (Palumbi et al., 2003). There is a ra pidly growing body of scientific research on the de sign of MPAs with biodiversity conservation as the primary planning objective (Les lie, 2005). However, Leslie ( 2005) indicates there is limited
5 research on designing networks of MPAs with connectivity and biodi versity conservation as concurrent planning objectives. Resear ch to determine how larval dispersal and oceanographic circulation can be used to evaluate potential connectivity among MPA sites has recently received increased attention, but is st ill very limited (Leslie, 2005; Palumbi et al., 2003). First and foremost, it is important to unde rstand the significance and dynamics of larval transport. Larval stages of marine organisms and th e transport strategies of their larvae are extremely complex and are a critical aspect of their population dynami cs. This realm of marine and spatial ecology requires multi-disciplinary effort and great expense to collect empirical data to even begin to understa nd marine larval biology and ecology. Even today we mostly rely on models and assump tions to understand the early life history of many marine organisms. What is known, is th at the early life hist ory of most marine benthic (occurring on the botto m) invertebrates and many fi sh involves a planktonic (passively floating and drifting) larval stage of development that acts as an agent for increased transport, dispersal, and gene flow between sessile (fixed) or sedentary and/or isolated adult populations. Passive planktonic larvae are at the mercy of ocean currents, winds, tides and other physical forces which determine their flow path, transport, and dispersal. Some marine species have larvae which begin as passively drifting, but then change into an actively swimming larvae stage. A combination of ocean current patterns and an actively swimming larval phase can limit the dispersal and tran sport of larvae over great distances, which enhances the pote ntial for self-seeding of certain marine populations.
6 One advantage of larval transport is that offspring are able to Â“escapeÂ” local environmental conditions (Gaines, 2005). Or ganisms without a planktonic larval stage (those with closed populations ) are not able to Â“escapeÂ” their local environmental conditions. Gaines (2005) stat es that there is strong eviden ce that species without larval transport and dispersal are more likely to be vulnerable to environm ental disturbance. There is also strong ev idence that species with open popula tions can be interdependent; if a critical source population is impacted by an unfavorable environmental disturbance, certain larval sink populations might decline due to the lack of recruitment. It is important to note empirical data on larval transport and conn ectivity of marine populations is very limited, resulting in consid erable debate as to the spatial scale and strength of larval connections betwee n populations (Mullineaux, DiBacco, Lerczak, Thorrold, Neubert, Caswell, Levin, & Largier, in preparation). Clearly, the transport of planktonic larvae in the marine environment is important to understand during MPA planning. Specifically those sessile organisms such as corals that are dependent on larval transport are at the center of marine conservation efforts to protect through the use of MPAs. MPAs are proven to be successful marine conservation and fishery management tools, but as t echnologies advance and more data become available, new MPA design strategies are co ntinually developed. With this in mind, it must be mentioned that even t oday there is considerable uncer tainty about the best spatial design of MPA networks (Largier, 2003). On a daily basis, MPA science is evolving and advancing in its endeavor to find the op timal MPA network design by understanding marine ecology better.
7 Largier (2003) emphasizes determination of larval transport distances and larval origins are a central challenge in contempor ary marine ecology. In the Caribbean region, Roberts (1997) suggests coral reefs that are supplied abundantly with larvae from Â“upstreamÂ” reef areas are likely to be more resilient to overfishing, less susceptible to species loss, and less reliant on local management than places with little Â“upstreamÂ” reef. With the goal of finding good techniques to exhibit Â“upstreamÂ” and Â“downstreamÂ” ecological links (i.e., larval c onnectivity) between MPAs, the intent of this thesis is to apply a spatial model of larval transport among coral habitats, Acropora palmata (Lamarck, 1816) populations, and MPAs with in the Florida Keys National Marine Sanctuary (FKNMS), and to examine patterns of connectivity among these areas using Geographical Information Systems (GIS). For the purpose of this thesis, it is as sumed the current MPAs in the FKNMS were designed with marine conservation as the pr imary planning objective, and the present research will assess connectivity among MPAs and coral habitats to aid managers in planning the addition of MPAs in the region in order to protect key coral populations based on their larval transport potential. The present research utilizes the combinat ion of GIS vector and raster analysis techniques to simulate larval transport and assess potential la rval connectivity. Environmental Systems Research Institute Â’s (ESRI) ArcGIS 9.1 and TarbotonÂ’s (1997, 2005) TauDEM and D flow routing are used to dete rmine potential larval transport paths, and assess MPA and co ral population connectivity. This thesis provides practical applica tion of connectivity theory using GIS; making possible a variety of spatial analysis options to evaluate potential larval
8 connectivity among MPAs and stony coral popula tions. Outcomes of the analyses will provide managers with an enhanced toolse t for planning and establishing networks of interdependent MPAs at local, regional and global scales. The next three sections of this chapter outline the research goal, objectives, and null hypotheses of this thesis, respectively. Th e final section of this chapter describes the organization of following chapters. Goal The research goal is to use a GIS-based model to describe the level of larval connectivity among coral habitats, A. palmata populations, and MPAs within the FKNMS. Objectives and Null Hypotheses Objective one. The first objective is to deve lop a GIS-based model of larval connectivity. Objective two. The second objective is to model th e level of larval connectivity among three A. palmata test populations and other co ral habitat within an 800 km2 study area in the Northeastern FKNMS. The null hypothesis is: Among the three A. palmata test populations, the mean August contributing fl ows from all other coral habitats are the same. Objective three. The third objective is to model th e level of larval connectivity among three A. palmata test populations and only other validated A. palmata populations.
9 The null hypothesis is: Among the three A. palmata test populations, the mean August contributing flows only from other validated A. palmata populations are the same. Objective four. The fourth objective is to compar e simulated larval connectivity among three A. palmata test populations with empirical genetic data. The null hypothesis is: Levels of larval connectivity do not have a positive relationship with clonal diversity among the three A. palmata test populations. Objective five. The fifth objective is to identify distant and unprot ected potential sources of coral larvae upstream of exis ting MPAs. The null hypothesis is: Mean August contributing flows from distant and unprotec ted coral habitats to existing MPAs are uniform throughout the study area. Objective six. The sixth objective is to descri be the potential sources of coral larvae upstream of existing MPAs. The null hypothesis is: Among different coral habitat types, the mean August contributin g flows to MPAs are the same. Chapter Organization The second chapter of this thesis is a literature review highlighting current knowledge that ultimately develops the theoreti cal framework for this research. The first and second sections describe the history of MPAs and the development of the National MPA Center, respectively. The third secti on outlines approaches to designing MPA networks. The fourth section describes th e importance of applying larval transport patterns to the design of MPAs. The fifth section describes current findings on the effectiveness of MPAs. The sixth section thoroughly descri bes theories of connectivity, with case examples of m easures of connectivity in spatial ecology, landscape
10 connectivity, the role of larval transport a nd dispersal, and connectivity in the marine environment. The last section describes in great detail the curre nt knowledge of MPAs, oceanography, larval transport, recruitmen t, and coral population within the FKNMS. The third chapter describes the met hodologies used for assessing larval connectivity among coral habitats, A. palmata populations, and MPAs. The models used to compute larval transport and levels of larval connec tivity among 1) three A. palmata test populations and all other coral habitats, including other validated A. palmata populations, and 2) coral habitats and MPAs are described. Methods for identifying unprotected distant coral habitats highly connected to MPAs in terms of larval transport are given. Details of how levels of connect ivity are statistically compared and mapped are described. The fourth chapter presents the results of the analyses. The levels of larval connectivity among 1) three A. palmata test populations and al l other coral habitats, including other validated A. palmata populations are examined and compared to population clonal structure. The levels of connectivity among coral habitats and MPAs are also examined, and unprotected sources of coral larvae for existing MPAs are mapped. The fifth chapter discusses the findings of this research. A review of the results and implications of the findings is presente d. A summary of contri butions and usefulness of this research are described. Finally, s uggestions for future research are presented.
11 Chapter Two: Literature Review History of MPAs According to Kendall Jr. and Picquelle (2003), Â“The 20th century was marked by increased exploitation of living marine resources and parallel increases in our attempts to manage these resources for long-term sustai nability.Â” Meester et al. (2004) expressed how: Â“The goals of policymakers for the worldÂ’s fisheries traditionally have been concerned with food production and employ ment.Â” We had gone from thinking the oceanÂ’s resources were unlimited, and available for uncontrolled exploitation, to trying to manage fisheries (Kendall Jr. & Picquelle, 2003 ). Attempts have been made to limit harvest, and even attempts to enhance them through hatcheries were made (Kendall Jr. & Picquelle, 2003). In spite of these management efforts, widespread overfishing occurred. Now, efforts to compensate for shortcomings of these resource management attempts, the creation of MPAs are increasingly gaining support (Davis, 1989; Bohnsack, 1993; Dugan & Davis, 1993), and have already been established in several places around the world (Wells & Keesing, 1990; Roberts & Polunin, 1992; Baker, Shepherd, & Edyvane, 1996; Airam, Dugan, Lafferty, Leslie McArdle, & Warner, 2003). MPAs include all areabased management efforts designated to enha nce conservation of marine resources or meet other objectives of ocean manage ment (National Research Council, 2001;
12 Lubchenco, Palumbi, Gaines, & Andelman, 2003; Leslie, 2005). In the United States, for example, the Marine Protected Areas Fede ral advisory Committee has identified 328 marine managed areas (Kendall Jr. & Picquelle, 2003). National MPA Center Executive Order No. 13158, signed in May of 2000, calls upon federal, state, local, and tribal governments and the private se ctor to work together to strengthen the protection of U.S. ocean and coastal resour ces (NMPAC, 2004). The order directed the National Oceanic and Atmospheric Admini stration (NOAA) to establish a National Marine Protected Areas Center (NMPAC) to provi de the science, tools, and strategies to help build a national system of MPAs (NMPAC, 2004). The specific objectives of the NMPAC (2004) are to provide resource managers with skills, products, and processes related to MPAs, and to develop products and services that can reduce duplicated efforts and increase efficiencies across a broad arra y of MPA efforts. According to NMPAC (2004), numerous decision-support tools, many of them GIS-based, have been created over the past few years to address a variety of issues both within and around MPAs. General Design of MPA Networks To put MPA science into perspective, Botsford, Micheli, and Hastings (2003) state: Â“The theory underlying the design of marine reserves, whether the goal is to preserve biodiversity or manage fisheries, is still in its in fancy.Â” The current status of MPA science is reviewed by Leslie (2005) and NMPAC (2004). NMPAC (2004) presents an inventory of GIS-based decisionsupport tools for MPAs. In list format, a
13 descriptive summary of each t ool explaining what the tool does, who developed it, what types of data are necessary to use it, if it is geographically specific, and how it may be useful to MPA activities is presente d by NMPAC (2004). A much more thorough synthesis of the use of these tools an d many other marine conservation planning approaches are presented by Les lie (2005). There has recently been an increasing interest in evaluating the effectiveness of marine c onservation and development projects (Leslie, 2005). With LeslieÂ’s (2005) evaluation of nume rous cases, the next step is to take what we have learned and develop standards fo r effective marine conservation. Some examples of these marine conservation planning approach es, specifically the planning and design of MPA networks will be de scribed in this literature review. Leslie (2005) discusses the effectivene ss of three main decision support tools: expert workshops, maps, and reserve selec tion algorithms. Leslie (2005) reviews how Groves (2003) provides a blueprint for the bringing together of people (in workshops) knowledgeable about the ecological, social, a nd economic aspects of the identified study region to guide planning for biodiversity c onservation. A prime example was how GIS maps and workshops were extremely valuab le tools in the pla nning of the Tortugas Ecological Reserve in the FKNMS (Franklin, 2002; Franklin, Ault, Smith, Luo, Meester, Diaz, Chiappone, Swanson, Miller, & B ohnsack, 2003; Cowie-Haskell & Delaney, 2003). Franklin (2002) discusse s how the process of planning and implementing of an MPA can be daunting, and that community a nd expert workshops and GIS maps were extremely effective tools in the planning a nd successful establishment of 2 MPAs known as the Tortugas Ecological Reserve in July of 2001. The planning effort was guided by community and expert based working gr oups that provided recommendations on the
14 preferred configuration of the Reserve. Franklin (2002) and Franklin et al. (2003) reveal how the utilization of GIS in the planning pr ocess provided several benefits. These key functions of GIS for MPA planning are presented by Franklin (2002): Â“(1) the preparation and display of ecological and so cioeconomic site char acterizations; (2) the functionality of intera ctive GIS to instantly query and upda te different scenarios at public forums and planning meetings; and (3) the advantage of using GIS to convey spatial relationships to stake-holders th rough enhanced imagery.Â” The third type of decisi on-support tools evaluated by Leslie (2005), computerbased heuristic and simulated annea ling algorithms (e.g., SPEXAN, SITES, and MARXAN), have proven useful in MPA design (Possingham et al., 2000; Airam et al., 2003; Leslie, Ruckelshaus, Ball, Andelman, & Possingham, 2003; Palumbi and Warner, 2003; Meester et al., 2004; Cook & Auster, 2005). Church et al. (2003) present results of a patch-building heuristic method, which should be very useful for conservation-reserve planning. The objective of using these simulatio ns is to generate various networks of potential protected or priority areas. For example, Meester et al., (2004) created multiple MPA plans and used a simulation model to asse ss the effects of reserve size and shape on select Florida Keys reef fish populations unde r dynamic spatial and temporal conditions. However, Meester et al. (2004) argued for a more comprehensive approach than using only one simulation model. Meester et al. (2004) proposed Â“an integrated sequence of simulation methodologies that provide an objective, quantitative framework for the design of marine reserves in a spatially heterogeneous coastal ocean environmentÂ”. According to Meester et al. (2004), these methodologies satisfy Â“the multiple, oftenconflicting criteria of disp arate resource user groupsÂ”
15 Applying Larval Transport Patterns to MPA Design One of the primary objectives of MPAs is to increase recruitment of target species both within the reserves and in adjacent areas ( Kendall Jr. & Picquelle, 2003). According to Kendall Jr. and Picquelle ( 2003), Â“the idea is that adults in MPAs which are free from harvest will live longer and grow larger, and since fecundity is directly related to fish size, roughly to length cubed, the larger fi sh will produce many more eggs.Â” The life cycle of most marine organisms has a disper sive planktonic life stage (Bohnsack, 1993). This suggests marine populations are Â‘openÂ’, with recruits to a population originating from adults elsewhere (Stobutzki, 2001). Afte r examining the early life history and larval transport distances of many marine orga nisms, Shanks, Grantham, & Carr, (2003) suggest MPAs be spaced far enough apart that long-distance dispersi ng larvae released from one MPA can settle in adjacent MPAs. Modeling larval transport and dispersal to aid MPA design is a fa irly new field of study, and is a very complex task. There ar e numerous unknown variables, and due to lack of data, assumptions are necessary. For ex ample, one must consid er that in order for recruitment enhancement to occu r, a fished area should be with in the transport distance of the eggs and larvae produced in an MPA (Gue nette et al., 1998; Botsford et al., 2001). For an MPA to act as a source for recruits to a fished area, prevailing currents must carry the eggs and larvae toward the fished area (D ahlgren et al., 2001). If currents run from the fished area to the MPA, the area could be considered a sink rather than a source of recruits, and would not enhance recruitment in the fished ar ea (Roberts, 1997; Crowder et al., 2000). Gerber, Botsford, Hastings, Po ssingham, Gaines, Palumbi, and Andelman (2003) state: Â“Although some m odels are beginning to yield information on the spatial
16 configurations of reserves required for popula tions with specific tr ansport distances to persist, it remains an aspect of reserve design in need of furthe r analysis.Â” Since little is known about larval transport a nd dispersal, networks of MP As which may act as sources of larvae are recommended (Roberts, Bohnsack, Gell, Hawkins, & Goodridge, 2001). Effectiveness of MPAs An evaluation by Halpern (2003) of over 100 studies of MPAs worldwide reveals that protection from fishing l eads to rapid increases in bi omass, abundance, and average size of exploited organisms and increased species diversity. Enhancing nearby fish populations is the most compelling aspect of MPAs for fishers and fisheries managers, although the effectiveness of this function is still under debate (Kendall Jr. & Picquelle, 2003). Although, Roberts et al. (2001) provide subs tantial evidence that MPAs in Florida and St. Lucia have enhanced nearby fisherie s. The authors argue that their results confirm theoretical predictions that MPAs can play a key role in supporting fisheries (Roberts et al., 2001). If this is accurate, then more fish will then be available for harvest in these adjacent areas that are open to fishing. Most marine fish have planktonic eggs, and along with the larvae are the primary transport and dispersal phases in fishes. It is suggested the eggs and larvae produced in a MPA will settle in the reserve and in adjacent areas to enhance recruitment both within the reserve and elsewhere (C arr & Reed, 1993; Kendall Jr. & Picquelle, 2003). However, Kendall Jr. and Picquelle (2003) discussed that in a review of 31 empirical studies on the effects of MPAs on target popul ations (both finfish and inve rtebrates), Dugan and Davis
17 (1993) found only three that considered recruitm ent effects: one of these showed positive effects and two did not demonstrate any effect. Connectivity Measures of connectivity in spatial ecology. According to Moilanen and Nieminen (2002), connectivity (or its inverse, isolation) is a fundamental concept widely used in spatial ecology to determine species distributions. Alt hough different ecological disciplines may use connectivity measures in slightly different contexts, metapopulation studies are concerned with in teractions between spatially distinct local populations (Moilanen & Nieminen, 2002). Moilanen and Nieminen (2002) primarily focused their study on connectivity measures in highly fr agmented environments (i.e. many habitat patches). In general, metapopulation studies typically use greatly simplified connectivity measures, such as distance to the neares t neighbor population, and the amount of habitat in a circle surrounding the ha bitat patch. However, Moil anen and Nieminen (2002) suggest that due to their extreme simplicity, it is questionable whether these measures are adequate in explaining phenomena related to th e spatial configuration of the habitat. Moilanen and Nieminen (2002) discuss a recent review by Tischendorf and Fahrig (2000) that discusses the defin ition, use, and misuse of the concept of connectivity. Tischendorf and Fahrig ( 2000) argue the appropriate measure of connectivity requires the measurement of actual immigration (or recruitm ent) rates. Here lies the challenge of modeling a complex and dynamic ecological relationship such as connectivity: Measurements of migration rates, even though important, are unfortunately very hard to come by (Moilanen & Niem inen, 2002). Tischendorf and Fahrig (2000)
18 summarize the current state of knowledge as fo llows: Â“Research is n eeded to determine what, if any, simple measures of landscap e structure can be used as measures of landscape connectivity.Â” Moilanen and Nieminen (2002) embarked on the task to investigate this issue by comparing several simple or relatively simple connectivity measures in their ability to predict colonization events in two large empirica l data sets on butterflies. They conclude that the simplicity of a nearest neighbor measure is not adequate. Buffer measures performed much better, but are sensitive to the size of the buffer. Results suggest that for highly fragmented habitats: Â“the best and most consistent performance is found for a measure that takes into account the size of the focal patch a nd the sizes of and distances to all potential source popul ationsÂ” (Moilanen & Nieminen, 2002). These measures of connectivity can be modeled many different ways. For example, these measures of landscape connectivity can be modele d using GIS or graph theory. Landscape connectivity. Landscape connectivity models have been built primarily on 2 types of spatial data, vector s (polygons) or raster grids (Urban, 2000). A less familiar approach, the use of the graph (Harary, 1969), in determining landscape connectivity using focal-species analysis in an island model has been demonstrated (Bunn, Urban, & Keitt, 2000; Cantwell & Fo rman, 1993; Halpin & Bunn, 2000; Urban & Keitt, 2001). Using a focal-species analysis, Bu nn et al. (2000) applied a graph-theoretic approach to landscape connectivity in the Coastal Plain of North Carolina. Bange and Hoefer (1976) presented a recent development at that time where various aspects of graph theory introduced powerful tools for geographers. According the Bange and Hoefer (1976), the best know n tool among geographers in the 1970s was
19 graph theory and its use in evaluating connect ivity of networks, accessi bility of locations, and other measures pioneered by Kansky ( 1963). Despite Bange and Hoefer (1976) being concerned with connectivity of a group of countries, their studies led to methods, thoughts, and ideas that later stimulated studies of habitat connectivity. The mathematical graph was used by Bunn et al. (2000) as an ecological construct with respect to habitat connectivity. They state, Â“Graph theory is a well established mainstay of information technologyÂ” (Bunn et al., 2000 ). According to Bunn et al. (2000) the graph is concerned with highly efficient ne twork flow, and can easily be adapted to landscape-level focal species an alysis. Bunn et al. (2000) were able to determine the functional distance between patches with a graph, which revealed the landscape was fundamentally connected for one focal species, but not for another. They argue the graph-theoretic approach is better than other modeling approaches because it can be applied with very little data and improved fr om the initial results. Urban and Keitt (2001) also demonstrate that a simple graph constr uct, the minimum spanning tree, can serve as a powerful guide to decisions about the re lative importance of individual patches to overall landscape connectivity. With an incr ease in GIS development, scientists have demonstrated the utility of GIS models to analyze landscape c onnectivity (Halpin & Bunn, 2000; Michels et al., 2001). A study by Michels, Cottenie, Neys, De Gelas, Coppin, & De Meester, (2001) demonstrates GIS modeling of the effec tive geographical dist ance among zooplankton populations in a set of inte rconnected ponds. Three GIS models were developed to simulate rates of zooplankton dispersal between ponds. Results indicate that the effective geographical distance as modele d by the flow rate and the di spersal rate model provide a
20 better approximation of true zooplankton dispersal than the Euclidian geographical distances or the landscape m odel that only considers the pr esence of physical connections (Michels et al., 2001). Halpin and Bunn (2000) utilized GIS to co mpute a least-cost distance matrix. This was a study comparing terrestrial and ma rine ecological applications of GIS to model connectivity (Halpin & Bunn, 2000). The authors ex plain that to assess the importance of individual pathways, a comple te set of possible paths must first be developed. In terrestrial situations, Halpin and Bunn (2000) describe how least-cost path algorithms can be used in an iterative manner to create a set of all potential paths between patches, resulting in a cost-distance matri x. Marine applications must consider directionality due to ocean currents to crea te the relative paths between patches. This requires two different types of path analysis approaches to deve lop the cost-distance matrix. Halpin and Bunn (2000) describe how with the terrestrial example, species traveling between patches are ex pected to move equally well in either direction, but this is not the case in their marine ex ample due to ocean current impedance. Larval transport and dispersal in the marine environment. Empirical data on larval transport and dispersal in the marine environment is limited. To fill this gap, there have been recent efforts to indirectly monito r species dispersal through chemical tags and genetic comparisons to help map populati on movements and measure the spread of species (Baums, Hughes, & Hellberg, 2005a; Baums, Miller, & Hellberg 2005b; Brazeau, Sammarco, & Gleason, 2005; Palumbi et al ., 2003). there are currently great interdisciplinary and collaborative efforts to Â“trackÂ” the early life history of several marine organisms, such as corals (Baums et al., 2005a, 2005b; Brazeau et al., 2005;
21 Sammarco, Atchison, & Boland, 2004), shrimp (Criales, Browder, Jackson, Robblee, & Hittle, 2003; Yeung et al., 2005), snappers (Jones, Lara, Yeung, Criales, Jackson, & Richards, 2005; Jones, Lara, & Lamkin, 2003) and bivalves (Becker, Fodrie, McMillan, & Levin, 2005; Mullineaux et al., in preparation) Since larval stages are microscopic, it is impossible to follow individuals, or to track them with conventional tags. With recent technological advances in DNA (Brazeau et al., 2005; Sammarco et al., 2004) and elemental (Mullineaux et al., in preparation) analyses, the evaluation of origins and trajectories of some plankt onic larvae is facilitated. For example, trace element fingerprinting by Mullineaux et al. (in preparation) determ ines the spatial scale and strength of connectivity among bivalve populat ions on the Massachusetts and southern California coasts. These chemical fingerprints or signatures in bivalves also allowed Becker et al. (2005) to determine the environmental conditio ns the larvae experienced during growth. This knowledge allowed reconstr uction of locations of larv ae. Becker et al. (2005) indicates that trace elemental fingerprinting is a promising technique to track bivalve larvae movement over long distances (up to 20 km). Becker et al. (2005) emphasize Â“Identification of spatial vari ation in elemental fingerprints that is stable over time represents a crucial step in enhancing our ability to understand larval transport and population connectivity in invertebrates.Â” This elemental tracking, in addition to advanced DNA tracking (Brazeau et al., 2005; Sammarco et al., 2004) are new tools that are beginning to shed light on many larval transport and dispersal mysteries, and will hopefully lead to groundbreaking discoveries in to the connectivity of populations. These
22 discoveries may also clarify the roles of phys ical, chemical and biol ogical processes that influence population distribution and abundance. It is very clear that determination of la rval transport and dispersal distances and larval origins is a major challenge in ma rine ecology (Largier, 2003). Largier (2003) focused on this problem from the perspectiv e of oceanography. Others have followed this approach also; for example, Thiba ut et al. (1998) highlights how hydrodynamic factors affect the recruitment of marine invertebrates in a ma crotidal area. It is also discussed by Kendall Jr. and Pi cquelle (2003) that through egg or larval transport (via ocean currents) some of the larvae will settle elsewhere and thus will enhance juvenile recruitment over an area much larger than the source itself (the Â“seeding effectÂ”). Todd (1998) addresses the issue of whether larvae always disperse as much as we believe. Todd (1998) demonstrates that even in highly dispersive environments with strong currents, certain benthi c invertebrates are behavioral ly constrained to minimize larval transport. The consequences of this discovery lead to the population being considered Â“closedÂ”. A population that wa s once thought to be Â“openÂ” is actually discovered to be Â“closedÂ”, thereby limiting population genetic differentiation. The lesson learned is to not make general deductions a bout Â‘opennessÂ’ of benthic assemblages based on a highly dispersive environment (Todd, 1998). Additional support for this conclusion is presented by Palumbi (1999), Swearer, Casell e, Lea, and Warner (1999), and Jones, Milicich, Emslie, and Lunow (1999). Palumbi (1999) reviews and discusses consequences of discoveries made by Swearer et al. (1999) and Jones et al. (1999). Understanding ocean current patterns is one of the major obstacles to biological ocea nographers (Palumbi, 1999). According to
23 Palumbi (1999), the basic assumption is that larvae drift the ocean s, traveling great distances and seldom returni ng to where they were spawned. Swearer et al. (1999) and Jones et al. (1999), each with different expe rimental approaches, demonstrate that the larvae of reef fish are not always disper sed great distances by st rong ocean currents. Palumbi (1999) states, Â“Instead, some are re tained near where they are spawned, and settle back onto the island reefs that their parents inhabitedÂ”. Th ese findings reveal the importance of understanding that larval transpor t and dispersal can vary greatly and is not always dependent on ocean circulation. Or maybe we only understand the tip of the iceberg when it comes to oceanography, and this is why we must eliminate assumptions by measuring ocean currents and lear ning early life histories better. Connectivity in the marine environment. The box on the left in Figure 1 illustrates all the potential scenarios for th e spatial connectedness of distant marine populations (i.e., that all populat ions are Â“openÂ” and dispersa l to all habitat patches is equal). When various factors are applied, th e number of possible scenarios dwindles. For example, in the marine environment, c onnectivity in relation to ocean currents and potential larval transport prev ents such openness and equal larv al flow as displayed in the box on the left. Also, a combination of variable s affecting larval transport, dispersal, and settlement impede such openness. In addi tion, the specific species and its reproductive mode play a big role in limiting or enhanci ng larval connectivity. In the box on the right in Figure 1, dominant ocean cu rrents during a particular or ganismÂ’s spawning season can dictate larval flow and potenti al larval connectiv ity if this organism has a passively drifting larval phase, there by highlighting which populations are potentially connected more than others. Ocean currents and species-specific reproductive modes (e.g., larval
24 transport strategy and spawning season) can drastically alter marine population connectivity. Figure 1. Potential Scenarios for the Spa tial Connectedness of Distant Populations Palumbi et al. (2003) reviews and discu sses how multiple methods and tools can help describe ecosystem patte rns over spatial and temporal scales that are directly relevant to conservation and ecosystem manage ment. Palumbi et al. (2003) describes the application of four new tool s being used in oceanography an d marine ecology to identify connectivity patterns and help design ocean re serves. Two of these tools, indirect monitoring of species disper sal through chemical tags a nd genetic comparisons, have already been reviewed in this chapter. Current knowledge on the 2 remaining applications, GIS and oceanography/ocean sens ing, will be reviewed in more detail. Â“Patterns of interconnection among marine resources have long been recognized as an important management concernÂ”, states Roberts (1997). It is possible to use ocean current patterns to identify connections among reefs. Roberts (1997) utilized surface current patterns to map transport routes of planktonic larvae from 18 coral reef sites in
25 the Caribbean. It was found that the sites varied, both as sour ces and recipients of larvae (Roberts, 1997). Results identified linkage s between sites Â“upstreamÂ” and Â“downstreamÂ” of each other, illustrating potential paths of gene flow for marine species with dispersive larvae. According to Roberts (1997), Â“The mapping of connectivity patterns will enable the identification of beneficial management partnerships among nati ons and the design of networks of interdependent reservesÂ”. A study currently underway by Kourafalou, Ba lotro, and Lee (2005) is the use of GIS and oceanography/ocean sensing to create an oceanographic model that represents the complex flow dynamics of the Southwest Fl orida shelf, Florida Keys and Florida Bay region. Â“The South Florida (SoFLA) Regiona l Model is an adaptation of the Hybrid Coordinate Ocean Model (HYCOM), hereaf ter called the SoFLA-HYCOMÂ” (Kourafalou et al., 2005). The SoFLA-HYCOM is a co mprehensive three-dimensional hydrodynamic ocean circulation model. Preliminary model validation with empirical ocean sensor data demonstrates reasonable agreement (Kourafalou et al., 2005). This model simulates the ocean current trends found throughout the re gion at different times of the year. Specifically, model results identify the different sized eddies or coastal countercurrents of the Keys that provide the larval pathways and opportunities for recruitment from both local and foreign s ources (Kourafalou et al., 2005; Lee, Williams, Johns, Wilson, & Smith, 2002). The SoFLA-HYCOM in combination with field measurements has helped delineate transport processes potentially linking South Florida Coastal ecosystems (Lee et al., 2002). The in corporation of these model computed ocean current patterns into a GIS-ba sed decision support system can aid in identifying potential
26 areas Â“upstreamÂ” and Â“downstreamÂ” of each other, highlighting potential interconnectedness of ecosystems. As mentioned previously in this chapter, different a pproaches to using GIS for measuring connectivity are required for terrestria l versus marine applications. Halpin and Bunn (2000) discuss how analysis of the potenti al connectivity of patchy marine habitats has become an important topic in marine conservation. Halpin and BunnÂ’s (2000) objective was to better unders tand the transport of planktoni c larvae from known habitat sites to other suitable habitat sites. R oberts (1997) conducted a generalized regional analysis to identify the amount of Â“ups treamÂ” and Â“downstreamÂ” reef area and approximate larvae travel time, but Halpin and Bunn (2000) argue little work has been done on developing spatial analysis tools for assessing connectivity within a reef system. To assess this problem, Halpin and Bunn ( 2000) used vector and raster analysis techniques in a GIS along with a physical oceanography model for the Mid-Atlantic and South Atlantic Bights to calculate larval flow paths and travel times among habitat patches. Results indicate that changes in current directions and velocities altered connectivity among the patches, requiring new habitat patch network solutions for each current regime in order to maintain connectivity (Hal pin & Bunn, 2000). Many assumptions are made when modeling connectivity. The old saying goes in this case: Â“Garbage in, garbage out.Â” Until th e appropriate amount of data is amassed to identify true connectivity of marine populations, we must rely on models which rely on significant assumptions. Assumptions about whether a marine population is open or closed, and the role of lo ng distance dispersal, are presented by Cowen, Lwiza, Sponaugle, Paris, and Olson ( 2000) and Warner and Cowen ( 2002). It is assumed most
27 marine populations are well connected via l ong-distance transport of larval stages (Cowen et al., 2000). Cowen et al. (2000) ex amined this assumption and found that when simple advection (transport by horizontal m ovement) models are used, larval exchange rates may be overestimated. According to Cowen et al. (2000), Â“such simplistic models fail to account for a decrease of up to nine or ders of magnitude in larval concentrations resulting from diffusion and mortalityÂ”. Th is indicates a marine population that was assumed open, is actually closed. Warner and Cowen (2002) took an additiona l analysis step: they incorporated realistic larval behavior and mortality estimat es and production variabil ity in their model. The results were consistent with their hypothesis that marine populations should be considered closed and must rely on mechanis ms enhancing self-recr uitment rather than depend on distant Â‘sourceÂ’ popul ations (Warner & Cowen, 2002). This finding is of great importance in the maintenance of marine population structures and management of coastal marine resources (Cowen et al ., 2000; Warner & Cowen, 2002). Florida Keys National Marine Sanctuary Background. The National Marine Sanctuary Pr ogram (NMSP) serves as the trustee for a system of 13 underwater sanctuaries and 1 co ral reef ecosystem reserve, encompassing over 150,000 square miles of marine and Great Lakes waters from Washington State to the Florid a Keys, and from Lake Huron to American Samoa (NMSP, 2005). Congress created the National Marine Sanctuary Program in 1972. The National Marine Sanctuaries Act (NMSA) authorizes the Secretary of Commerce to designate specific areas as National Marine Sanctuarie s to promote comprehensive management of
28 their special ecological, historic al, recreational, and aesthetic resources (Title 16, Chapter 32, Sections 1431 et seq. United States Code). Since the NMSA was enacted, it has been amended and reauthorized seven times. According to the NMSP (2005), Â“the amendments to the NMSA over the years have modified the process of how sites are designated, given the Secretary the authority to issue special use permits, enhanced the ability to enforce the Act, and established civil liability for injury to sanctuary resourcesÂ”. The National Oceanic and Atmospheric Ad ministrationÂ’s (NOAA), National Ocean Service (NOS) is responsible for manageme nt of the nation's Marine Sanctuaries. North America's only living coral barrier reef and the third longest barrier reef in the world (following Australia and Belize) lies about 10 km seaward of the Florida Keys (a 356 km island chain extending south and west of the Florida mainland), making it a unique national treasure of in ternational notoriety (FKNMS, 2005). These coral reefs are intimately linked to a marine ecosystem that supports one of the most unique and diverse assemblages of mangroves, seagrasses, har dbottom communities, patch reefs, and bankbarrier reefs in North Americ a (Cowie-Haskell & Delaney, 2003). Recently, significant degradation of the KeysÂ’ marine environment is the result, in part, of dramatic population growth throughout south Fl orida (USDOC, 1996). In an effort to address many complex thre ats to this important environment, to provide comprehensive protection to the region, and to ensure multiple, compatible use of resources, Congress created the Florida Ke ys National Marine Sanctuary (FKNMS) in 1990 (Florida Keys National Marine Sanctu ary and Protection Act, Pub. L. 101-605). The 9,800 square kilometer (km2) FKNMS surrounds the entir e archipelago of the Florida Keys and includes the productive wate rs of Florida Bay, the Gulf of Mexico and
29 the Atlantic Ocean. Recent ev idence of environmental decl ine within the sanctuary has created a sense of urgency to understand and protect the valuable resources within. Meester et al. (2004) emphasizes that the Florid a Keys are an ecosystem at risk as one of the nationÂ’s most significant, yet most stre ssed, marine resources under management of NOAA. MPAs. A comprehensive management plan for the FKNMS was adopted in 1997 that contained an innovative tool for marine resource protection, the creation of a network of 23 no-take zones, or MPAs: 18 small sanc tuary preservation areas, four special use areas and an ecological reserv e (FKNMS, 2005). The zones comprise less than 1 percent of the sanctuary, but protect much of its critical coral reef ha bitat. Effective July 2001, a second ecological reserve was cr eated in the Tortugas region, located in the westernmost reaches of the FKNMS (FKNMS, 2005). This Tortugas Ecological Reserve is divided into 2 sections, comprising 150 square nautical miles of ocean and includes the critical spawning grounds of RileyÂ’s Hump (USDOC, 20 00). The objectives of this reserve are to protect a full range of habita ts and preserve biodiversity. Studies clearly indicate that the Tortuga s region is unique in its location and the extent to which oceanographic processe s impact the area (USDOC, 2000). More importantly, the Tortugas plays a dynami c role in supporting marine ecosystems throughout south Florida and the Florida Keys (USDOC, 2000). Larvae that are spawned from adult populations in the Tortugas can be spread throughout the Keys and south Florida by a persistent system of currents and eddies that provide pathways necessary for successful recruitment (settlement) of both lo cal and foreign spawned recruits (juveniles)
30 with larval stages ranging from hours for so me coral species up to one year for spiny lobster (USDOC, 2000). Oceanography. After a 3 year collaborative e ffort, the Tortugas Ecological Reserve, the largest fully protected MPA in the U.S.A., was implemented in July 2001 (Cowie-Haskell & Delaney, 2003) Cowie-Haskell and Dela ney (2003) highlight how this process directly involved scientists and thei r input into the design of the MPA. Cowie-Haskell and Delaney (2003) describe how scientific information was derived, and how it influenced the siting and sizing of th e MPA. Overwhelming scientific research was committed to this purpose, and many groundbreaking discoveries into how this region is the oceanographic gateway to the entire FKNMS lead to a much improved understanding of largeand small-scale ocean circulati on patterns (Cowie-Haskell & Delaney, 2003; Lee, Johns, Wilson, & Williams, 1999; Lee & Williams, 1999; USDOC, 2000). Over 10 years of moored current meas urements, satellite-tracked drifters, shipboard hydrography and time sequences of satellite derived thermal images were analyzed (Lee, Clarke, Williams, Szmant, & Berger, 1994; Lee et al., 1999; Lee & Williams, 1999). Findings indicate the Tortug as region, located at the transition between the Gulf of Mexico and the A tlantic, is strongly influenced by 2 major current systems, the Loop Current in the eastern Gulf of Mexico and the Florid a Current in the Straits of Florida, as well as by the sy stem of eddies that form and travel along the boundary of these currents (Lee et al., 1994; Lee et al ., 1999; Lee & Williams, 1999; USDOC, 2000). Eddies are generally circular currents that run contrary to the main current. The formation of a large counter-clockwise rotating gyre (large eddy) that forms just south of
31 the Tortugas where the Loop Current turns abruptly into the Straits of Florida significantly influences marine communities of the FKNMS (USDOC, 2000). Lee et al. (1994) found that this gyre can persist for seve ral months before it is forced downstream along the Keys decreasing in size and increasi ng in forward speed until its demise in the middle Keys. This gyre serves as a retent ion mechanism for local recruits and as a pathway to inshore habitats for foreign recr uits (Lee et al., 1994; Lee & Williams, 1999). It may also serve as a potential food provider through plankton production and concentration (USDOC, 2000). Ocean circulation in the FKNMS is extr emely complex and dynamic. The most important aspect of circulation patterns is th at they favor the transport and retention of larvae and food throughout the entire regi on. A detailed description of how these dynamic current systems interact to favor marine communities throughout the FKNMS is given in the Tortugas Ecological Reserve final supplemental environmental impact statement and final supplemental management plan (USDOC, 2000). This document details how coastal curre nt systems create countercurrents which run primarily along the lower Keys and out to the Tortugas. According to USDOC (2000), the countercurrents provide a return route to the Tort ugas and its gyre-dominated circulation. In short, the e ffect of these currents on ma rine communities is to provide larval return mechanisms between the Tortugas and Florida Bay nursery grounds. Specifically, the complex combination of downs tream transport in the Florida Current, onshore Ekman transport (a process where by wind-driven upwelling bottom water is transported ~45 to the left of the actual wind direction in the northern hemisphere) along the coast, upstream flow in the coastal countercurrent, and recirculation in the Tortugas
32 gyre forms a recirculating recruitment pathwa y stretching from the Dry Tortugas to the middle Keys which enhances larval retenti on and recruitment into the Keys coastal waters (USDOC, 2000). The combination and variability of the different processes forming this Â“recruitment conveyorÂ” provide ample opportunity for lo cal recruitment of species with larval stages ranging from days to several months (Lee et al., 1994; Lee et al., 1999; Lee & Williams, 1999; USDOC, 2000). Larval transport and recruitment. Throughout the tropics, fish recruitment can occur over most of the year (Lindeman, P ugliese, Waugh, & Ault, 2000; Meester et al., 2004; USDOC, 2000). Colin, Sadovy, and Domeie r (2004) indicate spec ific conditions of biological cycles, physical oceanography a nd habitat tend to trigger fish spawning aggregations. For example, a number of sn apper spawning aggregation sites has been identified in the Tortugas region (Lindeman et al., 2000). These areas concentrate fish during the spawning season and serve as the source points for larvae that then drift passively and/or behaviorally (during a motile stage) until they become competent to metamorphose and settle to take on a bent hic existence (USDOC, 2000). Lindeman et al. (2000) highlights how commercial fishermen pr ovided evidence that groups of different species occupy different spawni ng sites at different times of the year. For example many snapper species ( Lutjanis sp. ) are thought to use the RileyÂ’ s Hump area as a spawning site (Domeier, 2004; Lindeman et al., 2000; USDOC, 2000). RileyÂ’s Hump is located approximately 10 nautical miles southwest of Dry Tortugas National Park (DRTO). This deep r eef terrace (22-27 m in depth) is not known for spectacular coral formations, but for its richness of fish and other marine life (USDOC, 2000). It is critical to protect the integrity of th e spawning sites and spawners
33 during the reproductive periods of the year, a nd to protect the habita ts critical to the survivorship of settling juve niles (USDOC, 2000). Under the fishery management plan (FMP) for reef fish developed by the Gulf of Mexico Fishery Management Council (GMFMC), RileyÂ’s Hump is closed May thr ough June to protect mutton snapper while they spawn (Lindeman et al., 2000). Lindeman et al. (2000) argue R ileyÂ’s Hump is the most important known snapper spawning aggreg ation site in the lower Florida Keys. Despite a 2 month site closure, aggregations of several other snapper species are heavily fished later in the year. Lindeman et al. (2000) believe a year-round closure to protect both fish stocks and remaining habitat integrity is warranted. Most tropical marine reef fishes have planktonic larvae that are dispersed by currents driven by winds, tides and bathymet ry. Recruitment of juveniles into a particular habitat or environment (e.g., the in shore coastal bays, near shore barrier islands or the coral reef tract) is de pendent upon the nature of the wa ter flow. Evidence of larval settlement of important reef fish species w ithin DRTO clearly exists (Lindeman et al., 2000). Interestingly, new evidence from physical oceanographers suggests gyre formations and current reversals occur seas onally which facilitate the transport and retention of larvae to suitable settling areas (USDOC, 2000). Migrations across the continental shelf are often nece ssary to connect settlement ar eas (sinks) to spawning sites (sources). Indeed, several sp awning sites in the Tortugas re gion have been identified by commercial fishermen and others (Lindeman et al., 2000). The probability of successful recruitment at a particular location is depe ndent upon the physical environment prevalent during the period of spawning and transport (USDOC, 2000). In general, the biophysical
34 processes involved in recruitm ent and survivorship of larvae is a very complex and dynamic stage of the life history of all marine organisms in the FKNMS. Stony Coral Populations. The Florida reef tract is the most extensive living coral reef system in North American waters and the third largest system in the world. All reefs are created by a community of reef-build ing organisms which produce calcium carbonate (CaCO3), providing the framework for organism s to inhabit. The primary reef-building organisms in the FKNMS are corals of the phyl ogenetic order Scleractinia. Scleractinian (stony) corals form the framework of some of the largest and most complex marine ecosystems on Earth, and these organisms form spatially structured populations (Mumby & Dytham, 2006) ideal for connectivity studies. According to Mumby and Dytham (2006) there is grave concern for the survival of stony coral populations worldw ide due to the imminent th reats from climate change (Hoegh-Guldberg, 1999) and other disturbanc es such as overfishing (Knowlton, 2001). Coral population connectivity is very dyna mic and difficult to grasp due to many variables such as predati on, disease, physical disturba nce, and overfishing (Mumby & Dytham, 2006). In addition, coral coloni zation is a complex multistage process combining production of offspring, transport, dispersal, arrival, settlement, and establishment (Mumby & Dytham, 2006). Baums et al. (2005b) used innovative t echnologies to iden tify two regionally isolated populations of the same species of A. palmata ; Western Caribbean and Eastern Caribbean metapopulations (with mixing in the central region near Puerto Rico) were found to be genetically differe ntiated. A metapopulation is a set of partially isolated populations belonging to the same species. The first analysis in the present study focuses
35 on examining larval connectivity and clonal diversity of documented A. palmata populations within the Western Cari bbean metapopulation in the FKNMS. Fadallah (1983) compiled knowledge and information on reproduction and development in stony corals a nd identified the sex, mode of reproducti on, type of larvae, timing of reproduction and planktonic larval duration (PLD) for 146 species throughout the world. All stony corals ge nerally display one of two se xually reproductive and larval transport strategies. These corals fall under either the Â“brood erÂ” or Â“broadcasterÂ” reproductive mode as described by Fadallah ( 1983) and Brazeau et al. (2005). These modes greatly differ in terms of fertiliza tion and larval phase. For example, brooder species display internal fertili zation of eggs and brood their la rvae before release into the water column, generally resul ting in a shorter PLD ranging fr om a few hours to days. Alternatively, broadcaster specie s demonstrate external fertil ization by releasing eggs and sperm into the water column simultaneousl y, resulting in a longer PLD ranging from a few days to months. Elkhorn coral (Acropora palmata) Within the FKNMS, the broadcaster A. palmata is a stony coral common throughout the Cari bbean and FL Keys (Figure 2). For millions of years, stony corals, including A. palmata reproduce in the middle of the night during just the right time in the lunar cycle by releasing eggs and sperm into the water column where they mix and fertilize (Baums et al., 2005a, 2005b). If all goes well, in as little as three days the Â“planulaÂ” Â– or coral babies Â– eventual ly find a suitable location to settle on the sea floor to coloni ze existing reef habitats or maybe even begin entirely new coral reefs. Most stony coral species spawn according to a l unar cycle, and, in the FL
36 Keys, spawning usually begins three to five days after the August full moon, about two hours after sunset (Baums et al., 2005a, 2005b). Acropora palmata (commonly referred to as Elkhor n Coral) has historically been the primary framework-building coral in the shallow Caribbean and FL Keys coral reef habitats. Its tendency to frag ment due to its delicate branches (Figure 2) allows it to rapidly proliferate resulting in monospeci fic and sometimes monoclonal colonies or stands. In spite of its ra pid growth and proliferation, A. palmata has undergone such widespread and drastic decline over the past 2 decades that it wa s recently listed as Threatened under the US Endangered Sp ecies Act (Federal Register, 2006). Figure 2. A Colony of A. palmata (Photo Courtesy of NOAA Center for Coastal Monitoring and AssessmentÂ’s Biogeography Team)
37 Species that build the physical structure of ecosystems like Acropora sp. often reproduce clonally (i.e., asexua lly) in addition to sexuall y (Figure 3). The degree of clonality or clonal diversity may vary over a speciesÂ’ range in accordance with the relative success of sexual and asexual recr uitment. High clonal diversity may promote species diversity and resilien ce in the face of environmen tal extremes. Conversely, low clonal diversity may indicate an asexual stra tegy to maintain resources during population decline. Figure 3. Flow Diagram of the Typical Acropora sp. Spawning Cycle (Photos Courtesy of www.undersea.com.au) www.undersea.com.au Asexual Â– Budding & Clonin g Sexual Â– Mass S p awnin g
38 Chapter Three: Methods and Materials Study Area The 9,800 km2 FKNMS surrounds the entire archip elago of the Florida Keys and includes the productive waters of Florida Bay, the Gulf of Mexico a nd the Atlantic Ocean (Figure 4). A study area within the F KNMS large enough to answer the research questions was chosen based on empirical data availability for A. palmata The present study focused on an 800 km2 area (Figure 5), from Boca Chita Key (northernmost) to Pigeon Key (southernmost). This Northeas tern section of the FKNMS extends from 24.5 N to 25.52 N and from 80.06 W to 81.16 W. The study area contains 15 no-take zones, or MPAs: 13 Sanctuary Preservation Areas (SPAs) and 2 Special Use or Research Only (SU) zones (Table 1). These MPAs protect a full range of habitats, including area s containing some of the sanctuaryÂ’s critical coral reef habitat. More importantly, this study area was chosen because it contains areas where recently collected empirical data on A. palmata population structure and genetics exist. Specifically, the location of three A. palmata test populations (Horseshoe, Little Grecian, and Sand Island Reef) and 25 km buffe rs of these populations (i.e., areas large enough to capture the long -distance transport of A. palmata larvae) identified the location and size of the study area illustrated in Figure 5.
39 Figure 4. Regional Map of the Florida Keys Na tional Marine Sanctuary (Courtesy of FKNMS) Florida Keys National Marine Sanctuary Marine Protected Areas Area To Be Avoided Ecological Reserves Existing Management Areas Florida Keys National Marine Sanctuary Florida State Waters John Pennekamp Coral Reef State Park National Park Boundaries National Wildlife Refuge Research Only Areas Sanctuary Preservation Areas Tortugas Bank No Anchoring Zone
40 810'0"W 8045'0"W 8030'0"W 8015'0"W 2430'0"N 2445'0"N 250'0"N 2515'0"N 2530'0"N Pigeon Key Boca Chita KeyHorseshoe Reef Sand Island Reef Little Grecian ReefFlorida Bay Atlantic Ocean Acropora palmata Populations MPAs Stony Coral Habitat Shoreline and Land Study Area Boundary 010203040 5 Kilometers Figure 5. The 800 km2 Study Area in Northeastern FKNMS
41 Table 1. Descriptions of MPAs within the Study Area MPA Name MPA Type Conch Reef Special Preservation Area Hen and Chickens Special Preservation Area Davis Reef Special Preservation Area Cheeca Rocks Special Preservation Area Alligator Reef Special Preservation Area Coffins Patch Special Preservation Area Sombrero Key Special Preservation Area Carysfort / South Carysfort Special Preservation Area Elbow Reef Special Preservation Area Key Largo Dry Rocks Special Preservation Area Grecian Rocks Special Preservation Area French Reef Special Preservation Area Molasses Reef Special Preservation Area Conch Reef (Research Only) Sp ecial Use / Research Only Tennessee Reef (Research Only) Special Use / Research Only Data Data incorporated into the analyses we re: the South Florida and Florida Keys shoreline, the FKNMS and MPA boundaries benthic (i.e., s ea floor) habitats encompassed by the boundaries of the FKNMS, empirical data on A. palmata population genetics, (i.e., population locations, genetic di versity, and spawning season), and a 1 year simulation of daily averaged three-dimensional ocean currents in the FKNMS region. Shoreline and boundaries. The South Florida and Flor ida Keys shoreline, and the FKNMS and MPA boundaries are polygon feature classes last updated in July of 2001, and were acquired from the Florida Fish and Wildlife Conservation CommissionÂ’s Fish and Wildlife Research Institute (FWRI) in St. Petersburg, Florida. The 1:40,000 scale shoreline (Figure 5) was digitized from NOAA nautical charts by FWRI. The FKNMS
42 and MPA boundaries (Figure 5) were digitized by FWRI based on th e legal description (i.e., bounding coordinates) in the Federal Register (2000). Benthic habitats. Benthic habitats are places on or near the sea floor where numerous aquatic organisms live, eat, and seek shelter (e.g., seagrass, mud, sand, hardbottom, coral reefs, etc.). The benthi c habitat resources of the FKNMS ecosystems have been extensively studied for several d ecades, creating a reliable long-term, system wide database for model parameterization (Meester et al., 2004). Precise mapping of these habitats in the FKNMS have enabled resource managers to make informed decisions about the protection of these re sources through the establishment of MPAs. The benthic habitats database used fo r the present study is the result of a cooperative effort between NOAAÂ’s National Ocean Service (NOS) and the FWRI to map the types and extent of benthic hab itats within the FKNMS (FWRI, 1998). The benthic habitats were mapped from a series of 450 aerial photographs of specific habitat types (24 are described) by interpreting co lor patterns on the photographs (FWRI, 1998). The types were classified in to 4 major categories: corals, seagrasses, hardbottom, and bare substrate. The habita t boundaries were georef erenced and digitized to create a polygon (vector) shapefile. This shapefile was last updated August of 1998. According to the metadata, horizontal accuracy of discrete points is with in 2 m, shoreline and reef habitats have an accuracy of 5 m, and seagrasses and other less resistant habita ts have an accuracy of 10 m (FWRI, 1998). Coral habitats. Stony (or scleractinian) corals are the most valuable and vulnerable natural resource throughout the F KNMS, which is why existing MPAs protect
43 several populations. Stony cora l populations were chosen as focal organisms for the present study, so their habitats were extrac ted from the benthic habitats dataset for analyses. All coral habitat polygons (Figur e 5) with the followi ng 10 benthic habitat descriptions were extracted from the benthi c habitats dataset: Patch Reefs Aggregated, Patch Reefs Aggregated with Halo, Patch Reefs Â– Coral or Rock Patches with Bare Sand, Patch Reefs Halo, Patch Reefs Indivi dual, Platform Margin Reefs Â– Back Reef, Platform Margin Reefs Â– Drowned Spur a nd Groove, Platform Margin Reefs Â– Reef Rubble, Platform Margin Reefs Â– Remnant Â– Low Profile, and Platform Margin Reefs Â– Shallow Spur and Groove. The majority of these habitat types are dominated by and/or built by stony corals. Acropora palmata populations. Using the coral habitat data extracted from the benthic habitats dataset relies on the key a ssumption that the assigned benthic habitat classes are an appropriate representation of coral biodiversity. These remotely sensed benthic habitats often lack the detailed info rmation about the distribution of species or population assemblages. This is why it is critical to utilize em pirical data on focal organisms, in addition to the bent hic habitat data mentioned above. Coral surveying efforts were optimized for A. palmata in 2006 (Miller, Chiappone, Rutten, & Swanson, in preparatio n). Results from Miller et al. (in preparation) provide the firstever baseline assessment of all A. palmata populations in the study area. Thirty four extant A. palmata populations in the study area were validated by Miller et al., (in preparat ion) and Baums et al. (2005b). The locations and validation dates of these populations, incl uding the Horseshoe (ID #12), Little Grecian (ID #8), and Sand Island Reef (ID #11) test popul ations are shown in Table 2.
44 Table 2. Locations of all Validated A. palmata Populations (Miller et al., in preparation) ID Location Description Reference Date Latitude () Longitude () 1 Elbow Reef Miller et al., (in pr eparation) 8/1/ 2001 25.1462 -80.2561 2 South Carysfort Reef Miller et al., (in preparation) 8/ 1/2001 25.2083 -80.2196 3 Carysfort Reef Miller et al., (in preparation) 8/1/ 2001 25.2216 -80.2099 4 NW of Conch Reef Miller et al., (i n preparation) 8/ 1/2001 24.9596 -80.4561 5 Pickles Reef Miller et al., (in preparation) 8/1/ 2001 24.9848 -80.4161 6 Molasses Reef Miller et al., (in preparation) 8/1/ 2001 25.0103 -80.3772 7 Sand Island Miller et al., (in pr eparation) 8/1/ 2001 25.0184 -80.3674 8 Little Grecian Reef Baums et al., (2005b) 8/17/2 003 25.1184 -80.3172 9 Boomerang Reef Baums et al., (2005b) 8/17/2003 25.3525 -80.1785 10 Marker 3 Baums et al., (2005b) 8/17/2003 25.3733 -80.1602 11 Sand Island Reef Baums et al., (2005b) 8/17/2003 25.0179 -80.3686 12 Horseshoe Reef Baums et al., (2005b) 8/17/2003 25.1395 -80.2944 13 North-North Dry Rocks Miller et al., (in preparation) 5/ 1/2005 25.1376 -80.2894 14 Sand Island Miller et al., (in pr eparation) 5/1/ 2005 25.0187 -80.3676 15 Key Largo Dry Rocks Miller et al., (i n preparation) 6/ 1/2005 25.1237 -80.2959 16 Carysfort Reef Miller et al., (in preparation) 6/1/ 2005 25.2229 -80.2094 17 Key Largo Dry Rocks Miller et al., (i n preparation) 6/ 1/2005 25.1249 -80.2981 18 Molasses Reef SPA Miller et al., (i n preparation) 8/ 8/2006 25.0092 -80.3748 19 Sand Island Miller et al., (in prep aration) 8/14/2 006 25.0183 -80.3684 20 French Reef SPA Miller et al., (in preparation) 8/17/ 2006 25.0356 -80.3477 21 Key Largo Dry Rocks SPA Miller et al., (in preparation) 8/ 21/2006 25.1233 -80.2976 22 NW of Elbow Reef SPA Miller et al., (in preparation) 8/ 21/2006 25.1682 -80.2699 23 Elbow Reef SPA Miller et al., (in preparation) 8/21/ 2006 25.1389 -80.2614 24 Elbow Reef SPA Miller et al., (in preparation) 8/21/ 2006 25.1412 -80.2596 25 Grecian Rocks SPA Miller et al., (in preparation) 8/21/ 2006 25.1093 -80.3059 26 Grecian Rocks SPA Miller et al., (in preparation) 8/21/ 2006 25.1105 -80.3040 27 North Dry Rocks Miller et al., (in preparation) 8/21/ 2006 25.1306 -80.2942 28 North-North Dry Rocks Miller et al., (i n preparation) 8/21/ 2006 25.1368 -80.2896 29 NW of Elbow Reef SPA Miller et al., (in preparation) 8/ 21/2006 25.1544 -80.2681 30 Near Maitland Grounding Site Miller et al., (in preparation) 8/ 21/2006 25.1974 -80.2268 31 Near Maitland Grounding Site Miller et al., (in preparation) 8/ 21/2006 25.1998 -80.2256 32 South Carysfort Reef SPA Miller et al., (in preparation) 8/ 21/2006 25.2075 -80.2224 33 South Carysfort Reef SPA Miller et al., (in preparation) 8/ 21/2006 25.2087 -80.2199 34 Southeast of Turtle Reef Miller et al ., (in preparation) 8/ 21/2006 25.2802 -80.2085 Acropora palmata population genetics. Genetic analyses of A. palmata populations by Baums et al. (2005a, 2005b, 2006) i ndicate three distant locations in the present study area show variab le levels of clonal diversit y. Baums et al. (2005a, 2005b, 2006) provide empirical evidence of low levels of genetic diversity within two of the three sampling locations. Specifically, clona l diversity of the Sa nd Island population was
45 significantly greater than clona l diversity of the other two populations at Horseshoe and Little Grecian Reef s. These three A. palmata populations at Sand Island Reef, Little Grecian Reef, and Horseshoe Reef (Figure 5 & Table 2), and their corresponding clonal diversity (Table 3) were se lected as Â“testÂ” populations for connectivity analyses. Table 3. Three A. palmata Test Populations and Their Clonal Diversity A. palmata Population Number of Cl ones Clonal Diversity Sand Island Reef 12 High (0.27) Little Grecian Reef 1 Low (1.00) Horseshoe Reef 1 Low (1.00) Modeled ocean currents. Modeled ocean currents from a comprehensive threedimensional hydrodynamic ocean circulation model for the Florida Keys, Southwest Florida Shelf and the shallow Florida Bay we re developed by Kouraf alou et al. (2005). This regional South Florida (SoFLA) model is an adaptation of the basin-scale HYbrid Coordinate Ocean Model (HYCOM). Accordin g to Kourafalou et al. (2005), nesting of the region-scale SoFLA-HYCOM model within a basin-scale model allows the accurate simulation of the interaction between sha llow water dynamics around the Florida Keys reef tract with basin-scale oceanic flows. The SoFLA-HYCOM area is shown in Figure 6, and it extends from approximately 22.6 N to 27.4 N (West Flor ida coast) and to 26.7 N (East Florida coast) and from 78.8 W to 83.8 W. The horiz ontal resolution is 1/25 degree (about 3 to 3.5 km in latitude) and 19 vertical circulat ion layers were implemented (i.e., the model incorporates bathymetry and topographic details to compute three-dimensional
46 circulation) with a 3 m minimum depth. The complex circulation dynamics of this region are adequately represented by this SoFLA-HY COM model, and results of the model were verified and found consistent with ocean drif ter, hydrographic surve y, and satellite data (Kourafalou et al., 2005). Figure 6. Geographic extent of the SoFLAHYCOM simulation (figure from Kourafalou et al., 2005). Dr. Vassiliki Kourafalou from the University of Miami, Rosentiel School of Marine and Atmospheric Science, provided raw data from a one year SoFLA-HYCOM simulation of georeferenced mean daily ocean cu rrent vectors, which were then translated into current direction angle (i n radians) through post-processi ng. The columnar text files provided contained latitude and longitude (decimal degrees ), along with corresponding U and V geostrophic current components (i.e., LLUV format).
47 Geostrophic refers to the balance between the Coriolis forces and the horizontal pressure forces. This balance produces a bala nced flow called a geostrophic current. The geostrophic current approximations are broken into its two horizontal components. The Â“UÂ” component (Ucomp) represents the eastwest component, while the Â“VÂ” component (Vcomp) represents the north-south compone nt. These components are oriented from True North at the locati ons of each vector. Ocean Current Data Post-Processing and Interpolation Point feature class creation. Many observations of A. palmata spawning events within the present study area i ndicate spawning occurs annual ly approximately three to five days after the full moon during the mont h of August. For this reason, each of the SoFLA-HYCOM daily text files during the month of August, to capture the most probable number of spawning days, was impor ted as tables into a Microsoft Access database. For each table, all of the record s with U and V component values equal to 9999 (these modeled points fall on land) were qu eried and deleted. The daily tables were added to an ArcMap document and the Add XY Data and Export Data tools were used to create point feature classes for each day. The map projection for the present study was a custom Florida Albers Conical Equal Area: False Easting = 400000, False Northing = 0, Central Meridian = -84, Standa rd Parallel = 24, Standard Pa rallel = 31.5, Central Parallel = 24, GCS = North American Datum of 1983 HARN. The map units were meters. Cross-validation of interpolated grids. Five percent of the points of one dayÂ’s current vector point feature class (Appendi x A; Figure A1) were randomly selected, extracted as a new feature class, and then de leted; these points in the new feature class
48 were used for cross-validation of three different interpolation grids. This resulted in two point feature classes; a 95% and a 5% featur e class which do not overlap and together they make up 100% of that dayÂ’s current v ector points. This cross-validation was necessary for evaluating which interpolation technique is be st at predicting the true SoFLA-HYCOM current vector values. To best represent the detail of inte rpolated SoFLA-HYCOM ocean current patterns and maintain reasonable data proces sing and storage require ments, a resolution of 300 m was chosen. Three grids using the 95% point feature cla ss as input points and the U component field as the Z value fiel d were calculated using three different interpolation techniques: Inve rse Distance Weighted (IDW), Spline (tension), and Krigin (default Krigin settings were maintained). These three U component interpolation gr ids were compared to points of known U component values (i.e., the 5% featur e class points) to determine how good the interpolation techniques predicted the U co mponent values at these locations (Appendix A; Figure A2). To do this the differences between actual and predic ted (interpolated) U component values were determined with z onal statistics (i.e., an overlay analysis) between the 5% feature class points and each of the three interpolation grids. For each point in the 5% feature class, the difference and absolute difference between the pointÂ’s known U component value and each of the thr ee interpolated values were calculated. The average of the difference and absolute difference were also calculated, and the result was Spline (tension) was best (Appendix A; Ta ble A1). The 95% a nd 5% feature classes were only created for cross-validation purposes; 100% of the current vector values from each daily averaged SoFLA-HYCOM simulation were input for the analyses.
49 Flow direction grid calculations. Figure 7 outlines the process of how the daily current vector points for each day in August we re further processed prior to analysis. Spline (tension) interpolations of the V and U component values using each August daily point feature class as input were perfor med. Spline interpolation was the favored technique based on the cross-va lidation of interpolated grid s results (Appendix A; Table A1). Using these U component (Ucomp) a nd V component (Vcomp) Spline interpolation grids as input, flow direction grids were co mputed for each of the 31 days in August. This was done by performing a combination of grid calculations using map algebra. Map algebra was used to correct the current vector angles so all points represent respective quadrants within a Cartesian plane. The corrected current vector angles and the angle theta were summed to get true angle measurement values (degrees) for each grid cell. The angle theta calculation was THETA = ABS (ATAN (Vcomp / Ucomp)) 57.2957795. The formula for current directi on in degrees was DIRECTION = CON (Ucomp > 0 & Vcomp > 0, 90 Â– THETA, Ucomp > 0 & Vcomp < 0, 90 + THETA, Ucomp < 0 & Vcomp > 0, 270 + THETA, Uc omp < 0 & Vcomp < 0, 270 Â– THETA. Finally, a DIRECTION (degrees) grid computati on is required to conve rt flow directions in degrees to an angle format (Figure 7) that the TauDEM D flow routing tools can use as input (Tarboton, 2005).
50 Figure 7. SoFLA-HYCOM Data Post-Processing
51 The TauDem D flow routing algorithm is ut ilized in the present study, and requires angle counter clockwise from East in radians as input (Figure 8). Map algebra was used to convert the DIRECTION (degrees ) grid to the TauDEM flow angle grid (Figure 7). First, the DIRECTION grids in degrees were converted to degrees counterclockwise from East (degCCfromE) with this calculation: degCCfromE = 90 Â– DIRECTION. Then the conditional statem ent CON (degCCfromE < 0, degCCfromE + 360, degCCfromE) was performed. Finally, the fl ow angle grids were calculated with the map algebra statement radCCfromE = deg CCfromE (3.14159 / 180). Flow directions are computed later in the analyses with the D flow routing algorithm (Figure 8); defined by the steepest planar slope on planar tr iangular facets on a block centered grid (Tarboton, 1997). Figure 8. D Method for Determination of Current Flow Directions (Figure from Tarboton, 2005)
52 The final post-processing results are flow direction (radians) grids (31 in total) for each daily averaged current regime in the month of August (Figure 7). These simulated SoFLA-HYCOM ocean current direction grids in addition to the data on coral habitats, existing MPAs, and A. palmata populations were inputs for the larval transport model and larval connectivity an alyses described below. A GIS-based Model of Larval Transport The Upslope Dependence Function of th e TauDEM 3.1 toolset for ArcGIS 9.1 was utilized to compute upstream depende nce grids for each SoFLA-HYCOM simulated daily averaged current regime. Specifically, fl ow direction (radians) and weight (target) grids were required input data for the TauDEM Upslope Dependence function, which uses the D flow routing algorithm (F igure 8) to compute the fraction of flow (at each grid cell) that contributes to any part of the targeted grid cells (Tarboton, 1997, 2005). This fraction of flow per grid cell simulate s the fraction of larvae flowing to the target cells. Basically, this function quantifies the amount (if any) a point x in the study area grid contributes to a point y in a targeted weight grid (Figure 9). Results from this upstream dependence function model the transpor t of larvae over a grid to target cells, which is useful for tracking where larvae may come from.
53 Figure 9. Upslope or Upstream Dependence Function of Grid Target Cells y (Figure from Tarboton, 2005) The simulated larvae in this model were assumed to reside in the mid-water column and to move passively with th e depth-averaged SoFLA-HYCOM simulated currents. Larvae were also assumed to m ove via simple advection (i.e., transport by horizontal movement). Due to these assumptions, this model is useful for organisms with a dominant planktonic larval phase and a ve ry short-term to no active swimming phase. All larval transport simulations and populat ion connectivity analyses were performed using the SoFLA-HYCOM daily averaged curren ts for the month of August, which is the prime spawning season of A. palmata and numerous other stony coral species (Baums et al., 2005a, 2005b).
54 Connectivity Analyses Numerous organisms throughout the Fl orida Keys coral reef ecosystems experience a planktonic larval phase during wh ich transport by ocean currents may occur. Knowledge of stony coral larval transport and settlement fr om distant populations in relation to ocean currents is limited. Warn er and Cowen (2002) addressed this problem by examining the role of long distance larv al transport versus local retention in replenishing marine populat ions. Recent coral genetics studies describe the clonal variation and potential larval connectivity of specific meta populations of stony corals in the Florida Keys region (Brazeau et al., 2005; Baums et al., 2005a, 2005b). Also, Shanks et al. (2003) highlights a tre nd where marine organisms display one of two dispersal strategies (i.e., short or long distance). Tw o GIS-based analytic modeling approaches to determining potential larval c onnectivity among coral habitats, A. palmata populations, and MPAs are described below and outlined in Table 4 and Figure 10. Table 4. Larval Source and Sink Areas in Analyses One and Two Connectivity Among: Analysis Larval Source Areas (x Grid Cells Within) Larval Sink Areas ( y Target Grid Cells Within) 1 Coral habitats (including those areas with validated A. palmata populations) Three A. palmata test populations: Little Grecian, Horseshoe, and Sand Island Reefs 2 Coral habitats MPAs
55 Figure 10. Modeling Overview (See Greater Detail in Appendix B)
56 Analysis One: Larval connectivity and A. palmata clonal diversity. The intent of this analysis was to mo del the level of larval connectivity among the three A. palmata test populations and all other coral ha bitat, including other validated A. palmata populations, and to compare this connec tivity to the clona l diversity of ea ch test population. The first objective of the following methodology was to model larval conn ectivity among each of the three A. palmata test populations and all other co ral habitats. The second objective was to model larval connectivity among each of the three A. palmata test populations and all other validated A. palmata populations. The final objectiv e was to compare variations in clonal diversity of each A. palmata test population with each test populationÂ’s simulated larval connectivity. First, the three A. palmata test population sites (Baums et al. 2005a, 2005b) were each converted to weight (or target) grid s with cell size set to 300 m, where the population site grid cell values were set to 1, and 0 elsewh ere. For each of the three target grids, the TauDEM Upslope Depende nce function was performed 31 times using each daily current direction (radians) grid fo r the month of August (Figure 9) as flow direction input (the proce ss is outlined in Figure 11). This resulted in upstream dependence grids for each A. palmata test population (i.e., target) and daily current regime in the month of August. Each output grid in Figure 11 simulates daily averaged larval transport in terms of contribu ting flow fraction per grid cell to each A. palmata test population. Batch processing to run the m odel 31 times (for each day and each test population) was used in the model illustrated in Figure 11. Next, larval connectivity among each A. palmata test population and all other coral habitat was determined for corals with long larval transport distances using zonal statistics.
57 Figure 11. Larval Trans port Model Ran for Each A. palmata Test Population and Each Daily Current Regime during August
58 Larvae of A. palmata were assumed to settle some where within 25 km from the spawning location; this assumption is supporte d by the evidence presented by Shanks et al. (2003). Based on the larval transport strategies highlighted by Shanks et al. (2003), a 0-25 km buffer around each of the three A. palmata test populations were calculated. These buffers should contain areas where bot h the larvae of long and short distance dispersing marine inverteb rate larvae, including A. palmata larvae, will settle. Each of these three larval transport buffer zones was intersected with the coral habitat polygons. These three intersections c ontained all coral habitats; including all 34 validated A. palmata population areas (Table 2), within 25 km of each A. palmata test population. The remaining steps of analysis one ar e highlighted in Figures 12A and 12B. Figures 12A and 12B depict one analytic pro cess divided into 2 figur es for visualization purposes (i.e., 12A flows into 12B; follow the flow arrows). Notice in these figures several input and output datasets are labeled with Â“PÂ”; these are steps in the analytic model that are model Parameters where batch processing takes place. Each label Â“PÂ” in the flow diagrams represents batch processi ng of each of the 31 daily contributing flow grids for each of the three A. palmata test populations. Some coral habitat polygons are ra ther small at approximately 100 m2 in total area. To perform an overlay analysis, or zonal statistics, where the contributing flow grids overlay the coral habitat polygon zones, gr id resolution must be decreased for flow values to be summarized for each coral habi tat polygon. To provide meaningful overlay statistical results for every coral habitat polygon zone, an Extract by Mask (Figure 12A) was necessary to lessen the cell size to 10 m fo r zonal statistics to be computed for every coral habitat polygon within 25 km of each test population (i.e, the zones). Zonal
59 statistics were computed for each A. palmata test populationÂ’s 31 extracted daily contributing flow grids (i.e., the value rasters). To enable data summaries and statistical analyses, a field was added and the day was calcu lated (to add the day as a table attribute) for each of the 31 daily zonal statistics output tables for each of the three test populations (Figure 12B).
60 Figure 12A. Flow Diagram of Analysis One
61 Figure 12B. Flow Diagram of Analysis One (Continued)
62 Zonal statistics allowed for modeling the level of larval connectivity. Specifically, the potential for each A. palmata test population to receiv e coral recruits via contributing flow from upstream coral reefs within 25 km (i.e., larval connectivity among each A. palmata population and all other coral hab itat) was quantified using zonal statistics, daily table summaries and table appends (Figures 12A and 12B). For each test population, daily contributing fl ow statistics from 1) all coral habitats, and 2) only validated A. palmata populations, was summarized. Statis tical analyses were performed on these daily larval connectivity statistics (n = 31) for each A. palmata test population. Statistical computations were performed with Analyse-It for Microsoft Excel. Descriptive statistics of da ily contributing flows by 1) all coral habitats, and 2) only validated A. palmata populations were computed. The Kolmogorov-Smirnov goodnessof-fit normality test (Moore, 1986), modified for use with unknown population mean and variance was performed on daily flows (n = 31) for each of the thre e test populations to verify parametric test assumptions. The non-parametric Kruskal-Wallis test wa s applied to test for a difference among the median contributing flows from 1) al l coral habitats, and 2) only validated A. palmata populations, to each A. palmata test population. AnalyseIt for Microsoft Excel computed the Kruskal-Wallis statistic describe d by Siegel and Catellan Jr. (1988). The pvalues were computed using the Chi-squa re approximation, with correction for ties (Siegel & Catellan Jr., 1988). The subse quent all-pairwise Mann-Whitney test was applied to indicate which pairs among the three A. palmata test populations are different. These tests allowed for determining if the mean ranks of contributing flow from 1) all coral habitats, and 2) only validated A. palmata populations, among the three A. palmata
63 test populations are similar; and if not, whic h population(s) differs from the other(s). These differences were then compared to vari ations in empirical genetic data documented for each test population by Baums et al. (2006). By using map algebra to add all of the 31 daily upstream dependence grids together for each target A. palmata test population, variable le vels of modeled larval connectivity among each test population and coral habitats were classi fied and visualized on a map. The coral habitats with high to tal August larval conn ectivity with each A. palmata test population were highlig hted on the map along with an overlay of the other validated A. palmata populations. These highlighted co ral habitat areas are potential larvae sources to the test populations, and may be considered hi gh priority sites for further genetic investigati ons to supplement those done by Baums et al. (2005a, 2005b; 2006). Analysis Two: MPA Larval connectiv ity and unprotected larvae sources. The intent of this analysis was to model the level of larval connectivity among MPAs and distant coral habitats, and to identify unprotected potential sources of coral larvae for future protection as MPAs. The first objective was to model le vels of larval connectivity among MPAs and distant coral habitat types. Determining which distant coral habitat types have the greatest and least downs tream influence on existing MPAs will aid managers in their decision-making and MP A planning processes. The second objective was to identify unprotected and distant co ral habitat areas co ntaining high larval connectivity with existing MPAs These unprotected and distant coral habitat areas are potential larval sources with the greatest downstream influence on existing MPAs, and
64 would be considered excellent MPA ca ndidates due to their high MPA larval connectivity. First, all 15 MPA polygons in the study ar ea were converted to one weight (or target) grid with cell size set to 300 m, where the MPA grid ce ll values were set to 1, and 0 elsewhere. Using this MPA grid as the target, the TauDEM Upslope Dependence function was performed 31 times using each daily current direction (radians) grid for the month of August (Figure 13). This resulted in upstream dependence grids for all MPAs (i.e., the targets) and each daily current regime in the month of August (each output grid simulates daily averaged larval transport in terms of contributing flow fraction per grid cell to any MPA). Next, daily averaged la rval connectivity among the MPAs and distant coral habitats was determined for corals with long larval transport distances using zonal statistics. A flow diagram of this analys is is illustrated in Figure 13. This figure illustrates the process for modeling daily la rval connectivity among MPAs and distant coral habitats, and identifying unprotected a nd distant potential sources of stony coral larvae. The Â“PÂ” represents stages in the model where batch processing was used to run the model 31 times; once for each daily ocean current regime and MPA upstream dependence grid (Figure 13).
65 Figure 13. Flow Diagram of Analysis Two
66 Coral habitats within 3 km buffer zones of MPAs were not considered in this analysis since Shanks et al. (2003) states this may be a se lf-seeding zone. Yes, it would make sense to expand the borders of current MPAs if bordering areas have high larval contribution potential, but the purpose of th is methodology was to identify distant coral habitat areas not currently near or within an MPA with the greatest downstream influence (i.e., high potential to serve as larvae sources) on existing MPAs. To do this, a polygon feature class of 0Â–3 km MPA buffers was created (Figure 13). The MPA 3 km buffer polygons were used to erase coral habitats within 3 km of existing MPAs. The resulting polygon feature class contained only those coral habitats > 3 km from any MPA; this was used as the ma sk feature class to ex tract each of the 31 MPA upstream dependence grids. The extracted grid cell size was set to 10 m to run zonal statistics on all of these coral hab itat polygons > 3 km from any MPA (i.e., the zones), using each of the 31 extracted upstream dependence grids as value rasters. This resulted in 31 contributing flow zonal statistics output tables (Figure 13). Day fields were added and each daily va lue (01-31) was calculated in each of these daily contributing flow tables (Figure 13). Summary statistics and table appends of each of the 31 contributing flow tables allo wed for the creation of a daily summary of mean contributing flow to any MPA by coral habitat type (Figure 13). Statistical computations were performed with Analyse-It for Microsoft Excel. Descriptive daily contributing flows by coral ha bitat type statistics were computed. The Kolmogorov-Smirnov goodness-of-fit normality te st (Moore, 1986), modified for use with unknown population mean and variance was performed to verify parametric test assumptions.
67 The non-parametric Kruskal-Wallis test wa s applied to test for a difference among the median contributing flows from the 10 cora l habitat types to MPAs. Analyse-It for Microsoft Excel computed the Kruskal-Wallis statistic described by Siegel & Catellan Jr. (1988). The p-values were computed us ing the Chi-square approximation, with correction for ties (Siegel & Catellan Jr ., 1988). The subsequent all-pairwise MannWhitney test was applied to indicate which pa irs of contributing co ral habitat types are different. These tests allowed for determining if the mean ranks of contributing flow to MPAs among the 10 coral habitat types were similar; and if not, which coral habitat type(s) differs from the other(s). By using map algebra to add all of th e 31 daily MPA upstream dependence grids together, variable levels of modeled la rval connectivity among MPAs and distant unprotected coral habitats were classified and visualized on a map. Levels of August larval connectivity among distan t (> 3 km from any MPA) unpr otected coral habitats and MPAs were classified and mapped. Visually interpreted regions of contiguous coral habitats with high contributing flows to MPAs were extracted and zonal statistics of contributing flow per region were performed. These coral habitat ar eas are identified as potential larvae sources to MP As, and may be considered MPA candidates due to their high levels of larval connec tivity with downstream MPAs du ring the month of August. The protection of these areas may benefit st ony coral species diversity by preserving gene flow potential among populations duri ng their predominant spawning season.
68 Summary of Assumptions This thesis relied on models and a ssumptions to interpret and understand a complex phenomenon that occurs during the early life history of many marine organisms. There are numerous unknown and unmeasured vari ables, and due to lack of certain data, some assumptions were necessar y. Below is a list of this studyÂ’s assumptions critical to be aware of when considering th e implications of the results. 1) It was assumed the SoFLA-HYCOM ocean current simulations with an approximate 3 km resolution were suffici ent to meet the analysis goals. The SoFLA-HYCOM is the newest and highest resolution ocean circulation model available for the study area, and there is substantial evidence the simulations have been validated and are considered to accu rately represent ocean currents in the study area. 2) It was assumed A. palmata larval transport occurs by simple advection (transport by horizontal movement) since the domina nt planktonic larval stage passively drifts within the water column and transport is highly dependent upon ocean currents. The biological implications of passively drifting larval stages are increases in the chance of long distan ce transport and inte r-connectedness of distant populations. However, actively sw imming larval stages can inhibit long distance transport and increas e the likelihood of larval retention and self-seeding of local populations. It is unknown how th e more minor (i.e. of short duration) actively swimming larval stage of A. palmata impacts transport via ocean currents. It was considered a reasonab le assumption within the scope of this thesis to consider the larvae passi vely drifting by simple advection.
69 3) It is assumed the D flow routing algorithm represents the flow of larvae in an acceptable manner to meet analysis goals. The implications of this assumption are that horizontal and verti cal mixing of larvae was not considered; this is due to flow routing limitations. The D flow routing is deterministic and water always flows in a fixed predictable fashion; the water is only allowed to flow horizontally in 2 directions and never back into a cell water previously flowed from. The flow of water is not perfect fo r representing the mixing that occurs naturally in the ocean, but with present software and mode ling capabilities and for the scope of this thesis, it is acceptable to represent larval transport with D flow routing. 4) It was assumed each daily upstream dependence grid (n = 31) was a unique dailyaveraged snapshot or obse rvation of the potential for a pool of larvae released anywhere in the study area to enter the targeted area, regardless of time. These upstream dependence grids do not model larv al transport over one day or over any time period. They do simulate for each daily-averaged ocean current regime, larval transport regardless of time if th e currents remained constant. In other words, they are 31 independent unique observa tions of simulated larval transport. 5) It was assumed the flow fraction per cell (i.e., contributing flow) represented the fraction of a larvae pool (if any) entering the ta rgeted cell(s). This assumption does not take into account predation or other currently unknown factors reducing larval survivability and/or recruitment rate; the biological implication being an over-estimate of larvae entering the targeted cell(s), or larval transfer rate. 6) It was assumed the 34 validated A. palmata populations represented all extant A. palmata populations within the study area, a nd they were considered equal when
70 modeling larval connectivity. A populat ionÂ’s colony size and surface area influences the potential for contributing larvae, but these data are not available at this time for all 34 populations. The bi ological implications of this assumption are over and under-estimates of the potent ial for each of these populations to contribute larvae. For this thesis, treati ng each population with equal potential to contribute larvae was acceptable due to the lack of empirical data. 7) Based on the larval transport strategies identified by Shanks et al. (2003), it was assumed larvae of many broadcastin g stony coral species, including A. palmata settle on the reef within 3 km to 25 km of the spawning location. 8) It was assumed larval connectivity was the estimate of larval transfer from upstream populations to targeted downstr eam populations; it was measured in terms of contributing flow. Levels of larval connectivity among populations are assumed to identify and quantify ecol ogical relationships or the interconnectedness of distant populations, allowing for the understanding of the potential for genetic exchange among popul ations in relation to ocean currents during the month of August.
71 Chapter Four: Results Analysis One: Larval Connectivit y and A. palmata Clonal Diversity Modeled larval transport. Grids of daily contributing flow during the month of August to each targeted A. palmata test population were comput ed; these grids predict the daily averaged fraction of fl ow (from 0.0 to 1.0) for every grid cell contributing to a downstream A. palmata test populationÂ’s targeted cell. The daily grids were totaled for each A. palmata test population to visualize total August snapshots of contributing flow and flow fractions were converted to per cent values (Figures 16, 17, and 18). These figures display levels of water flowing to the target during August in percent flow per grid cell. Ocean rivers flowing into each ta rget can be visualized in Figures 16, 17, and 18 (Symbology Stretch Type: Standard Deviations, n: 1, cell size: 300 m).
72 Figure 14. Total August C ontributing Flow Grid for the Sand Island Reef A. palmata Test Population
73 Figure 15. Total August Cont ributing Flow Grid for the Little Grecian Reef A. palmata Test Population
74 Figure 16. Total August C ontributing Flow Grid for the Horseshoe Reef A. palmata Test Population Modeled larval connectivity. The basic concept of this analysis is to infer larval transport in relation to domi nant ocean currents (Figur e 17) varies among the three A. palmata test populations. Figure 17 illustrate s the SoFLA-HYCOM simulated ocean current direction varies thr oughout the study area. This section describes how the results indicate these variable ocean currents and the distribution of potentia l larval sources (i.e.,
75 coral habitats) throughout the study area influenc e levels of larval connectivity among the three A. palmata test populations. Figure 17. One Daily Averaged SoFLA-HYCOM Simulation of Ocean Current Direction in Relation to Coral Habitat and A. palmata Test Population Locations Contributing flows from all coral habitats, including other validated A. palmata population areas, within 25 km buffers of each A. palmata test population (Figure 18) were determined. This is based on the assu mption coral larvae set tle within 25 km of
76 their parentÂ’s location. The co ral habitat polygons were used as a mask to extract each A. palmata test populationÂ’s daily contributing flow grids. The extracted daily grids were totaled for each A. palmata test population to visuali ze total August snapshots of contributing flow from coral habitats with in 25 km (Figures 20, 21, and 22). These figures visualize total August larval c onnectivity among coral habitats and each A. palmata test population (Symbology Stretch Type: St andard Deviations, n: 1, cell size: 10 m). They are total August snapshots of wher e coral larvae potentially come from during the month of August. The predicted levels of contributing larv ae are flow fractions converted to percent flow per cell.
77 Figure 18. Coral Habitats and Validated A. palmata Populations
78 Figure 19. Total August Larval Connectivity among Coral Habitats and the Sand Island Reef A. palmata Test Population
79 Figure 20. Total August Larval Connectivity among Coral Habitats and the Little Grecian Reef A. palmata Test Population
80 Figure 21. Total August Larval Connectivity among Coral Habitats and the Horseshoe Reef A. palmata Test Population Statistical analyses. Daily levels of cont ributing flow among each A. palmata test population from all other coral habitats with in 25 km for the month of August were determined (Appendix C, Table C1). These co mputed mean flow fr actions represent the predicted levels of daily larval connec tivity among each test population and all other coral habitats within 25 km. Tests for normality indicate the distribution of daily contributing flow values among the A. palmata test populations do not meet parametric
81 test assumptions. The daily contributing fl ow values are not normally distributed (Kolmogorov-Smirnov test, p < 0.05), although the shapes of the distributions among A. palmata test populations are similar. Figure 22 shows box-plots of mean daily (n = 31) contributing flows (flow fractions) from all coral habitats (within 25 km) to each A. palmata test population. This figure allows for comparison of differences among the three test populations. The boxplots show the central locati on and scatter/dispersion of me an daily contributing flow fractions by test population (computed by An alyse-It for Microsoft Excel). See Appendix D (Figure D1) for a descripti on of the box-plot symbols. Mean August (total of 31 days) contributing flow from all coral habitats within 25 km to the Sand Island Reef population is 0.025 SE 0.0027 (Figure 22 and Table 5). Mean August contributing flow fr om all coral habitats within 25 km to the Little Grecian Reef population is 0.006 SE 0.0007 (Figure 22 and Table 5). Mean August contributing flow from all coral habitats w ithin 25 km to the Ho rseshoe Reef population is 0.003 SE 0.0004 (Figure 22 and Table 5).
82 -0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 Sand IslandLittle GrecianHorseshoe Figure 22. Box-Plots of Mean Daily Contribut ing Flows from All Coral Habitats within 25 km to Each A. palmata Test Population Table 5. Summary Statistics of August Contri buting Flow from All Coral Habitats to Each A. palmata Test Population Mean Contributing Flow from All Coral Habitat by Population n Mean SD SE 95% CI of Mean Sand Island 31 0.025 0.0148 0.002 7 0.020 to 0.031 Little Grecian 31 0.006 0.0041 0.000 7 0.005 to 0.008 Horseshoe 31 0.003 0.0022 0.000 4 0.002 to 0.004 Daily levels of contributing fl ow only from other validated A. palmata populations within 25 km among each A. palmata test population were determined for the month of August (Appendix E, Table E1). These computed mean flow fractions represent the predicted levels of daily la rval connectivity among each test population and Flow Fractions
83 other A. palmata populations within 25 km. Tests for nor mality indicate the distribution of daily contributing flow values among the A. palmata test populations do not meet parametric test assumptions. The daily contributing flow values are not normally distributed (Kolmogorov-Smirnov test, p < 0.05), although the shapes of the distributions among A. palmata test populations are similar. Mean August (total of 31 days) contributing flow from all validated A. palmata populations within 25 km to the Sand Island Reef population is 0.116 SE 0.0018 (Figure 23 and Table 6). Mean August contributing flow from all validated A. palmata populations within 25 km to the Li ttle Grecian Reef population is 0.037 SE 0.0020 (Figure 23 and Table 6). Mean August contributing flow from all validated A. palmata populations within 25 km to the Horseshoe Reef population is 0.033 SE 0.0007 (Figure 23 and Table 6).
84 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 Sand IslandLittle GrecianHorseshoe Figure 23. Box-Plots of Mean Daily Contri buting Flows Only from Other Validated A. palmata Populations within 25 km to Each A. palmata Test Population Table 6. Summary Statistics of August Cont ributing Flow Only from Other Validated A. palmata Populations to Each A. palmata Test Population Contributing Flow from Validated A. palmata Populations n Mean SD SE 95% CI of Mean Sand Island 31 0.116 0.0098 0.0018 0.112 to 0.119 Little Grecian 31 0.037 0.0109 0.0020 0.033 to 0.041 Horseshoe 31 0.033 0.0036 0.0007 0.032 to 0.034 The non-parametric one-way Kruskal-Wallis test and the all-pairwise comparisons Mann-Whitney test were performed. The mean ranks of August contributing flow from all coral habitats within 25 km significantly differ among the Flow Fractions
85 three A. palmata test populations (Kruskal-Wallis test, p < 0.0001, Table 7). The Sand Island test population received the highest mean rank of contributing flow from all coral habitats within 25 km (Table 7). The Little Grecian and Horseshoe Reef test populations received lower mean ranks of contributing fl ow from all coral habitats within 25 km (Table 7). An all-pairwi se comparison indicates each A. palmata test populationÂ’s mean rank of August contributing flow from all coral habitats within 25 km significantly differed from each other (Mann-Wh itney test, p < 0.0001, Table 7). Table 7. Kruskal-Wallis Test and Subseque nt Mann-Whitney All-Pairwise Comparison Test for Differences in Larval Connectivity among Each A. palmata Test Population and All Other Coral Habitats Mean Contributing Flow from All Coral Habitat N = 93 Kruskal-Wallis testn Rank sum Mean rank Sand Island 31 2325.0 75.00 Little Grecian 31 1332.0 42.97 Horseshoe 31 714.0 23.03 Kruskal-Wallis statistic 58.50 p <0.0001 (chisqr approximation) Mann-Whitney test2-tailed p Sand Island v Little Grecian <0.0001 (normal approximation) Sand Island v Horseshoe <0.0001 (normal approximation) Little Grecian v Horseshoe <0.0001 (normal approximation) Mean ranks of August contributi ng flow only from other valid A. palmata populations within 25 km signi ficantly differ among the three A. palmata test populations (Kruskal-Wallis test, p < 0.0001, Table 8). Th e Sand Island test population received the highest mean rank of contributing flow from other valid A. palmata populations within 25 km (Table 8). The Little Grecian and Hors eshoe Reef test popul ations received lower
86 mean ranks of contributing flow from other valid A. palmata populations within 25 km (Table 8). When comparing the mean ranks of Augus t contributing flow from other valid A. palmata populations within 25 km, mean ranks significantly differed among the Sand Island and Little Grecian Reef test populat ions (Mann-Whitney test, p < 0.0001, Table 8), and among the Sand Island and Horseshoe Reef test populations (Mann-Whitney test, p < 0.0001, Table 8. Mean ranks were similar among the Little Grecian and Horseshoe Reef test populations (Mann-Whitney test, p = 0.4974, Table 8). Table 8. Kruskal-Wallis Test and Subseque nt Mann-Whitney All-Pairwise Comparison Test for Differences in Larval Connectivity among Each A. palmata Test Population and Other Validated A. palmata Populations Contributing Flow from A. palmata Populations N = 93 Kruskal-Wallis testn Rank sum Mean rank Sand Island 31 2418.0 78.00 Little Grecian 31 929.0 29.97 Horseshoe 31 1024.0 33.03 Kruskal-Wallis statistic 62.07 p <0.0001 (chisqr approximation, corrected for ties) Mann-Whitney test 2-tailed p Sand Island v Little Grecian <0.0001 (normal approximation) Sand Island v Horseshoe <0.0001 (normal approximation, corrected for ties) Little Grecian v Horseshoe 0.4974 (normal approximation,corrected for ties) Baums et al. (2006) provided empirical genetic data on the clonal diversity of each A. palmata test population (Table 9). Clonal diversity in the Sand Island population was significantly greater than the clonal diversity of the other two populations at Horseshoe and Little Grecian Reefs (Baums et al., 2006). Based on empirical data on
87 number of colonies sampled, number of clone s, and clonal diversit y, the Sand Island Reef population was classified as mostly sexual, and the Horseshoe and Little Grecian Reef populations were classified as asexual (Bau ms et al., 2006). Table 9 clearly highlights the positive relationship between mean ranks of larval connectivity and clonal diversity among the three test populations. Table 9. Comparison of Differences in Larval Connectivity (Kruskal-Wallis mean ranks) and Clonal Population Structure among the Three A. palmata Test Populations Acropora palmata Test Population Little Grecian Horseshoe Sand Island Mean Rank of Connectivity With Other A. palmata Populations 29.97 33.03 78.00 Mean Rank of Connectivity With All Coral Habitats 42.97 23.03 75.00 Number of Clones (B aums et al., 2006) 1.00 1.00 12.00 Clonal Diversity (Baums et al., 2006) Low (1.00) Low (1.00) High (0.27) Reproductive Classification (Baums et al., 2006) Asexual Asexual Sexual Mean ranks of contributing flow, wh ether it is only from other valid A. palmata populations or from all other co ral habitats within 25 km, are significantly higher for the Sand Island test population when compared to th e Little Grecian and Horseshoe Reef test populations (Tables 8, 9, and 10). Analysis Two: Larval Connectivit y and Unprotected Larvae Sources Modeled larval transport. Grids of daily contributing flow during the month of August to all targeted MPAs (Figure 24) we re computed; these grids predict the daily averaged fraction of flow (from 0.0 to 1.0) for every grid cell contributing to any downstream MPAÂ’s targeted cell. The daily grids were totaled to visualize a total August
88 snapshot of contributing flow to all MPAs (F igure 25) and flow fractions were converted to percent values (Symbology: Manu al Classification, cell size: 300 m). Figure 24. Marine Protected Areas within the Study Area
89 Figure 25. Total August Contri buting Flow to all MPAs Modeled MPA larval connectivity. Daily contributing flow to all MPAs from all coral habitats greater than 3 km from any MPA was determined. Coral habitats contributing flow to MPAs within 3 km of any MPA were omitted from this analysis to eliminate any MPA self-seeding effects. More specifically, when trying to identify unprotected areas highly connected to existing MPAs, those areas greater than 3 km from
90 any MPA are given the highest priority for futu re protection due to their vulnerability or great distance (>3 km) from any pr otected larval s ource or sink. The coral habitat polygons gr eater than 3 km from any MPA were used as a mask to extract each MPA daily cont ributing flow grid. The extract ed daily grids were totaled to visualize a total A ugust snapshot of contributing flow from coral habitats greater than 3 km to any MPA (Figure 26). Figure 26 visu alizes total August larval connectivity among coral habitats and MPAs. This map visu alizes levels of contributing flow from distant (> 3 km from any MPA) unprotected coral habitats to MPAs (Symbology: Manual Classification, cell size: 10 m). It is a total August snap shot of where coral larvae potentially come from during the month of A ugust. The predicted le vels of contributing larvae are flow fractions converted to percent flow per cell.
91 Figure 26. Total August Larval Connec tivity among Coral Habitats and MPAs Statistical analyses. All coral habitats within 3 km from any MPA were excluded from the following analysis results. Daily le vels of contributing flow from all 10 coral habitat types (Figure 27) to MPAs for the month of August were determined (Appendix F, Table F1 ). These computed mean flow fractions represent the predicted levels of daily larval connectivity among MPAs and each co ral habitat type greater than 3 km from any MPA.
92 Figure 27. Ten Coral Habitat Types within the Study Area Mean August contributing flow to MPAs from each coral habitat type is summarized in Figure 28 and Table 10. The greatest mean contributing flow to MPAs (0.344 SE 0.0272) comes from the habitat type Â“Patch Reefs Â– Coral or Rock Patches with Bare SandÂ” (Table 10). Other coral ha bitat types with high contributing flow to MPAs (Figure 28 and Table 10) are Â“Platfor m Margin Reefs Â– Remnant Â– Low ProfileÂ”
93 (0.303 SE 0.0240), Â“Platform Margin Reefs Â– Drowned Spur and GrooveÂ” (0.260 SE 0.0227) and Â“Patch Reefs Â– IndividualÂ” (0.225 SE 0.0152).
94 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Patch Reefs Aggregated Patch Reefs Aggregated with Halo Patch Reefs Coral or Rock Patches with Bare Sand Patch Reefs Halo Patch Reefs Individual Platform Margin Reefs Back Reef Platform Margin Reefs Drowned Spur and Groove Platform Margin Reefs Reef Rubble Platform Margin Reefs Remnant Low Profile Platform Margin Reefs Shallow Spur and Groove Figure 28. Box-Plots of Mean Daily Contribut ing Flow from Unprotected and Distant Cora l Habitats to MPAs, by Coral Habitat Typ eFlow Fractions
95 The total areas of each unprotected and distant (> 3 km from any MPA) coral habitat types are summarized in Table 11 a nd Figure 29. The Â“Platform Margin Reefs Â– Back ReefÂ” and Â“Patch Reefs Â– HaloÂ” coral habitat types ha ve the lowest contributing areas (242,400 m2 and 355,800 m2, respectively) and cont ributing flows (0.100 and 0.083, respectively) to MPAs (Tables 11 and 12). However, the Â“Patch Reefs IndividualÂ” coral habitat type despite having the fourth hi ghest mean contributing flow (0.225) had substantially lo wer contributing area (2,038,700 m2) than 6 other habitat types; indicating cont ributing area does not have a str ong relationship with contributing flow (Figure 29). Table 11. Summary Statistics of August Contributing Area from Distant Coral Habitats to MPAs, by Coral Habitat Type Coral Habitat Type Total Area (m2) Platform Margin Reefs Back Reef 242,400 Patch Reefs Halo 355,800 Platform Margin Reefs Shallow Spur and Groove 522,500 Patch Reefs Individual 2,038,700 Platform Margin Reefs Reef Rubble 3,064,800 Patch Reefs Aggregated with Halo 6,708,700 Patch Reefs Aggregated 7,172,800 Patch Reefs Coral or Rock Patches with Bare Sand 12,769,700 Platform Margin Reefs Remnant Low Profile 19,528,100 Platform Margin Reefs Drowned Spur and Groove 24,892,600 Table 10. Summary Statistics of August Contributing Flow from Distant Coral Habitats to MPAs, by Coral Habitat Type Contributing Flow to MPAs by Habitat Type n Mean SD SE 95% CI of Mean Patch Reefs Aggregated 31 0.121 0.0704 0.0127 0.096 to 0.147 Patch Reefs Aggregated with Halo 31 0.121 0.0704 0.0127 0.096 to 0.147 Patch Reefs Coral or Rock Patches with Bare Sand 31 0.344 0.1514 0.0272 0.288 to 0.399 Patch Reefs Halo 31 0.083 0.0551 0.0099 0.063 to 0.103 Patch Reefs Individual 31 0.225 0.0845 0.0152 0.194 to 0.256 Platform Margin Reefs Back Reef 31 0.100 0.0637 0.0114 0.077 to 0.123 Platform Margin Reefs Drowned Spur and Groove 31 0.260 0.1264 0.0227 0.214 to 0.306 Platform Margin Reefs Reef Ru bble 31 0.167 0.0775 0.0139 0.138 to 0.195 Platform Margin Reefs Remnant Lo w Profile 31 0.303 0.1338 0.0240 0.254 to 0.353 Platform Margin Reefs Shallow Spur and Groove 31 0.180 0.1112 0.0200 0.140 to 0.221
96 0 5 10 15 20 25 30Pat c h Reefs Halo Platf o rm M a rgin Reefs Back R eef Pa t c h Re efs Agg r eg at e d wit h H a l o Pa t ch Reefs Ag g re g at e d Plat f orm Margin R e efs Reef Ru b ble Plat for m Mar g i n R e e f s Sha l l o w Sp ur and Gr o ove Pat c h Reefs Indivi d ual Pl a t f orm Margin Re ef s D rown e d S pur a nd Gr oov e Pla t for m Margi n Re e f s Remn ant L ow Pr ofi l e Patch Reefs Co r al o r Rock Pa t ches with Bare Sa nd0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 Total Area (km2) Contributing Flow to MPAs (Mean Flow Fraction) Figure 29. Contributing Area (km2) and Contributing Flow to MPAs (Mean Fl ow Fraction) among Coral Habitat Types Total Contributing Area to MPAs (km2) Mean Contributing Flow to MPAs (Mean Flow Fraction) km2 Mean Flow Fraction
97 Mean ranks of August contributing flow to MPAs significantly differ among the 10 coral habitat types (Kruskal-Wallis test p < 0.0001, Table 12). An all-pairwise comparison of mean ranks of August contribu ting flow to MPAs am ong the coral habitat types indicates several significant differences (Mann-Whitney test, p < 0.05, Appendix G, Table G1). The coral habitat type Â“Patch Reefs Â– Cora l or Rock Patches with Bare SandÂ” had a significantly greater mean rank of August contributing flow to MPAs (Kruskal-Wallis mean rank = 242.34) than 8 other coral ha bitat types (Mann-Whitney test, p < 0.05, Appendix G, Table G1). Only one coral ha bitat type had a mean rank similar to the Â“Patch Reefs Â– Coral or Rock Patches with Bare SandÂ” t ype; and that was Â“Platform Margin Reefs Â– Remnant Â– Low ProfileÂ” (Mann-Whitney test, p = 0.2342, Appendix G, Table G1). Table 12. Kruskal-Wallis Test for Differences in Coral Larval Connectivity with MPAs among Coral Habitat Types Kruskal-Wallis test N = 310 Contributing Flow to MPAs by Habitat Type n Rank sum Mean rank Patch Reefs Aggregated 31 3421.5 110.37 Patch Reefs Aggregated with Halo 31 3421.5 110.37 Patch Reefs Coral or Rock Patches with Bare Sand 31 7513.0 242.35 Patch Reefs Halo 31 2371.0 76.48 Patch Reefs Individual 31 5830.0 188.06 Platform Margin Reefs Back Reef 31 2791.0 90.03 Platform Margin Reefs Drowned Spur and Groove 31 6412.0 206.84 Platform Margin Reefs Reef Rubble 31 4552.0 146.84 Platform Margin Reefs Remnant Low Profile 31 7083.0 228.48 Platform Margin Reefs Shallow Spur and Groove 31 4810.0 155.16 Kruskal-Wallis statistic 120.56 p <0.0001 (chisqr approximation, corrected for ties)
98 Unprotected larvae sources. Visual identification of areas with high modeled larval connectivity among distan t unprotected coral habitats and MPAs was performed. The map from Figure 30 was visually examined for the largest areas of contiguous coral habitat with high mean Augus t contributing flow to MPAs (i.e., large coral habitat patches with flow greater than 40% per grid cell). Six regions that visually met these contiguousnesses and contributing flow parame ters were roughly outlined (Figure 30). Figure 30. Unprotected and Distant Coral Habitat Regi ons with High Total August Larval Connectivity with MPAs
99 Six regions of contiguous coral habitat with high contributing flow to MPAs (> 0.4 mean flow fraction per habitat type) we re visually interpreted (Figure 30) and described (Table 13). Regions 6 and 1 (Figure 30) had the greater mean August connectivity with MPAs among the regions and coral habitat types (T able 13). Despite regions 6 and 1 having high connectivity with MPAs, they had the lowest (Region 6: 1,300 m2) and third lowest (Region 1: 2,600 m2) total contributing coral habitat areas (Table 13). The region 6 mean contributing flow to MPAs (0.6605 SD 0.0197) was slightly higher than the mean flow from region 1 (0.6565 0.1298). Mean contributing flows to MPAs from the remaining four re gions in descending order were region 2 (0.5308 0.0158), region 5 (0.5289 0.0337), region 3 (0.4835 0.0522), and region 4 (0.4791 0.0270). The Â“Platform Margin Reefs Â– Remnant Â– Low ProfileÂ” was the dominant coral habitat type with the greatest total Augus t contributing area within region 1 (1,100 m2), region 3 (5,000 m2), region 4 (1,700 m2), region 5 (2,200 m2), and region 6 (500 m2). Region 2 was dominated by the Â“Patch R eefs Â– AggregatedÂ” and Â“Patch Reefs Â– IndividualÂ” coral habitat types; with contributing areas of 3,000 m2 and 1,700 m2, respectively (Table 13). The Â“Platform Margin Reefs Â– Remnant Â– Low ProfileÂ” and Â“Patch Reefs Â– Coral or Rock Patches with Ba re SandÂ” coral habitat types were the only two types with contributing flow to MPAs w ithin every region. Regions 2 and 3 contain the greatest total contributing area (5,900 m2 and 5,700 m2, respectively) to MPAs. Region 2 contains high mean August contributing flow to MPAs (0.5308 SD 0.0158), the greatest total Augus t contributing area (5,900 m2), and is the only region containing an unprotected and distant valid A. palmata population (Figure 30, and Table 13).
100 Table 13. Summary Statistics of Larval C onnectivity and Contributing Area (by habitat type) among Each Region containing High To tal August Larval Conn ectivity with MPAs Region Coral Habitat Type Area (m2) Mean Contributing Flow to MPAs (flow fraction) STD 1 Patch Reefs Aggregated 600 0.7830 0.0777 1 Platform Margin Reefs Remnant Low Profile 1,100 0.6360 0.2116 1 Patch Reefs Individual 800 0.6275 0.2299 1 Patch Reefs Coral or Rock Patches with Bare Sand 100 0.5793 0.0000 Total 2,600 2 Patch Reefs Coral or Rock Patches with Bare Sand 100 0.6934 0.0000 2 Platform Margin Reefs Drowned Spur and Groove 200 0.6440 0.0350 2 Platform Margin Reefs Shallow Spur and Groove 200 0.5720 0.0173 2 Platform Margin Reefs Back Reef 100 0.5292 0.0000 2 Platform Margin Reefs Reef Rubble 100 0.5041 0.0000 2 Platform Margin Reefs Remnant Low Profile 500 0.4537 0.0491 2 Patch Reefs Individual 1,700 0.4275 0.0138 2 Patch Reefs Aggregated 3,000 0.4225 0.0108 Total 5,900 3 Platform Margin Reefs Drowned Spur and Groove 200 0.5468 0.0205 3 Patch Reefs Coral or Rock Patches with Bare Sand 500 0.4608 0.0462 3 Platform Margin Reefs Remnant Low Profile 5,000 0.4430 0.0899 Total 5,700 4 Platform Margin Reefs Drowned Spur and Groove 100 0.5557 0.0000 4 Patch Reefs Coral or Rock Patches with Bare Sand 500 0.4873 0.0291 4 Platform Margin Reefs Reef Rubble 200 0.4682 0.0099 4 Platform Margin Reefs Remnant Low Profile 1,700 0.4051 0.0691 Total 2,500 5 Platform Margin Reefs Remnant Low Profile 2,200 0.5397 0.0451 5 Patch Reefs Coral or Rock Patches with Bare Sand 300 0.5324 0.0395 5 Platform Margin Reefs Reef Rubble 300 0.5276 0.0182 5 Platform Margin Reefs Drowned Spur and Groove 200 0.5160 0.0320 Total 3,000 6 Platform Margin Reefs Reef Rubble 100 0.6931 0.0000 6 Patch Reefs Coral or Rock Patches with Bare Sand 300 0.6618 0.0332 6 Platform Margin Reefs Drowned Spur and Groove 400 0.6517 0.0141 6 Platform Margin Reefs Remnant Low Profile 500 0.6353 0.0315 Total 1,300
101 Chapter Five: Discussion and Conclusions Introduction Mora and Sale (2002) define connectivity as the measure of the rates of exchange of individuals among populations, and fo r most marine organisms, population connectivity is largely driven by the processes th at influence larval transport. The results of the present study modeled the transfer of coral larvae among near by or more distant, local coral populations and MPAs There exists very limited knowledge on this type of larval connectivity for any coral reef or ganism (Sale, 2006), yet the present studyÂ’s methodologies and results are a cr ucial step in the right direct ion if we are to improve our ability to design and implement networks of MPAs in spatial arrangements that preserve and/or enhance marine population connectivity. The present methodologies strongly relie d on modeling and GIS. Modeling larval transport with the TauDEM Up slope Dependence function and D flow routing algorithm using the SoFLA-HYCOM simulated ocean current vectors, coral habitats, and MPAs as input allowed the computation of c ontributing flows. Contributing flows from coral habitats to other cora l habitats or MPAs simulate d larval connectivity. Data summaries allowed for the analysis and visual interpretation of tre nds and differences. All of these methodologies have the common element of deriving new maps and datasets of the likely occurrence or ma gnitude of larval connectiv ity based on an established
102 relation between existing GIS data. This is why modeling lies at the very core of analytical applications in GIS (Eastman, 2001). The present study demonstrated the fi rst use of ArcGIS and TauDEM to successfully model how major ocean current s during August create potential larval transport paths that may enhance gene fl ow via larval connectivity among coral populations and MPAs within the upper and middle Florida Keys. Specifically, this study first modeled the fr action of water flowing from any grid cell in the study area, to any downstream targeted grid cell during every day in August. Four targets were used for quantifying this type of larval tranport: the Horseshoe Reef, Little Grecian, and Sand Island Reef A. palmata test populations (Analysis One), and all MPAs (Analysis Two). This model simulate d the movement of the water mass per grid cell in which larvae would travel within. This study then quantifie d levels of larval connectivity among upstream coral habitats and each targeted grid cell. Results provided evidence major ocean currents during August may impede larval transport among coral populations in certain regions of the study area. The present results reveal this biophysical process signi ficantly influences la rval connectivity among A. palmata populations, MPAs, and coral habitats. These significant differences in larval connectivity may explain the significant variations in clonal diversity of the three A. palmata test populations documented by Baums et al. (2006). The following discussion highlights how th e results of this study successfully modeled levels of larval conn ectivity and determined signifi cant variations in larval connectivity occur during A ugust throughout the study area. Understanding where coral larval recruits may come from is the very foundation of learning the dynamics of larval
103 connectivity, and this study reveals potential sources of coral larvae for MPAs. These results and methodologies will drastically e nhance our ability to design and implement networks of MPAs in spatial arrangement s that preserve and/or enhance marine population connectivity. Larval Connectivity and Clonal Divers ity of A. palmata Populations Among the three A. palmata test populations, mean August contributing flows from 1) all other coral habitats, and 2) only from other validated A. palmata populations, are not similar. The difference in larval connectivity among the three test populations is highly significant, allowing for the rejection of the null hypotheses in research objectives two and three. This evidence supports the alternative hypothesis th at levels of August larval connectivity among each of the three A. palmata test populations and other coral habitats, including only other validated A. palmata populations vary. One surprising conclusion is that levels of larval connectivity have a positive relationship with clonal divers ity among the test populations, a llowing for the rejection of the null hypothesis in research objective four. Among the three A. palmata test populations, larval connectivity was significantly greater be tween the clonally diverse Sand Island Reef population and upstream cora l habitats. Conversely, the monoclonal Horseshoe and Little Grecian Reef A palmata populations had signifi cantly less larval connectivity with upstream coral habitats, including other validated A. palmata populations. These findings indicate the locatio ns of habitats in re lation to major ocean current patterns during August may greatly influe nce the rates of exchange of coral larvae among populations. This is based on the assumption that varied A. palmata clonal
104 diversity is due to varied levels of larv al inflow from other populations. The exact explanation is unknown, but it is a very surpri sing positive relationship this study reveals between larval connectivity and clonal di versity, justifying further investigation. Critical Unprotected Coral Habitat Upstream of Existing MPAs Mean August contributing flows from di stant and unprotected coral habitats to MPAs are not similar among habitat types. The difference in larval connectivity among coral habitats and MPAs is highly signifi cant among habitat types, allowing for the rejection of the null hypothesis in research objective six. Re sults support the alternative hypothesis that levels of August larval conne ctivity among each of the ten coral habitat types and MPAs vary significantly. Three coral habitat types are significan tly more connected to downstream MPAs compared to the remaining seven habitat t ypes. Throughout the st udy area, Â“Patch Reefs Â– Coral or Rock Patches with Bare SandÂ” Â“Platform Margin Reefs Â– Remnant Â– Low ProfileÂ”, and Â“Platform Margin Reefs Â– Dr owned Spur and GrooveÂ” habitat types have significantly more downstream influence on MPAs during the month of August. The results identify those coral habitat type s with the greatest and least influence on downstream MPAs. With this knowledge, habi tat composition should be a consideration when designing additional MPAs due the vari ability of larval c onnectivity among coral habitat types and existing MPAs. After visualizing larval c onnectivity among individual co ral habitats and MPAs, it is obvious mean August contributing flows from distant and unprotecte d coral habitats to existing MPAs are not similar throughout the st udy area, allowing for the rejection of the
105 null hypothesis in research objective five. This evidence supports the alternative hypothesis that levels of August larval conne ctivity among individual coral habitats and MPAs vary significantly throughout the study area. Results re veal that during the prime spawning season for multiple coral species, di stant and unprotected coral habitats with downstream influence on existing MPAs are diffe rentiated from those habitats with very little to no downstream influence. A map of coral habita t regions with high contributi ng flows to MPAs allowed for the identification of critical unprotected coral hab itat upstream of existing MPAs, including one region contai ning the single distant (> 3 km from any MPA) and unprotected validated A. palmata population in the study area. Due to the obviously high larval connectivity between this region and the presence of the threatened species A. palmata this region should be immediately protec ted and declared an MPA. This would help protect the present A. palmata population directly, and it w ould indirectly benefit the highly connected downstream habitats and MP As by protecting their larval sources; especially the currently unprotected A. palmata larval source. The remaining regions of high connectiv ity with MPAs are excellent candidates for further study and possible protection as MPAs The results of this study compliment each other in the effort of finding MPA larval sources and planning new MPAs by providing the knowledge that it is not only critical to finding unprotected co ral habitats highly connected to MPAs, but to also analyz e the composition of the coral habitat types within these areas. Certain coral habitat types are signifi cantly disconnected from MPAs in the study area. Evaluating a region with high connectivity and high contributing area from the habitat types that ar e underserved throughout the st udy area (i.e. contribute little
106 flow to existing MPAs) would be a more holistic approach to studying, identifying, and ultimately preserving potential coral larvae sources. It is clear now that spec ies distribution (i.e., coral ha bitat type) in addition to spatial optimization of MPAs with respect to larval connectivity s hould be taken into consideration when planning an MPA network or adding new MPAs within the FKNMS. A recent modeling approach by Matisziw and Murray (2006) illustrated how spatial associations and spatial distribut ions of reserves affect long-te rm persistence of species. The present results clearly indi cate larval connectivity vari es over space w ithin the study area. Evidence is mounting that larval connec tivity should be considered, in additional to species abundance and distribution, wh en designing MPA networks. Summary of Contributions Below is a list of the contributions of new knowledge that this thesis makes. 1) Developed and demonstrated the first use of a GIS-based model of larval transport to A. palmata populations and MPAs using ArcGIS and TauDEM. 2) Developed and demonstrated the first use of a GIS-based model of larval connectivity among A. palmata populations, coral habi tats, and MPAs using ArcGIS and TauDEM. 3) Determined simulated levels of larval connectivity among each A. palmata population at Horseshoe, Little Grecian, and Sand Island Reefs and other coral habitats significantly differ.
107 4) Determined simulated levels of larval connectivity among each A. palmata population at Horseshoe, Little Grec ian, and Sand Island Reefs and other A. palmata populations significantly differ. 5) Determined simulated levels of la rval connectivity among each of the Horseshoe, Little Grecia n, and Sand Island Reef A. palmata populations may be a determinant of clonal diversity. 6) Determined simulated levels of larval connectivity among coral habitats and MPAs significantly vary among coral habitat type. 7) Identified distant and unpr otected potential sources of coral larvae upstream of existing MPAs. Summary of Limitations and Assumptions It is important to recognize this thes is relied on models and assumptions to interpret and understand a complex phenomenon th at occurs during th e early life history of many marine organisms. Assumptions due to data a nd software limitations were acceptable for the present analysis objectiv es, but further analyses would benefit by overcoming some of these limitations. The SoFLA-HYCOM simulations are appr oximately 3 km in resolution, which limits the analysis because the results are only meaningful at this resolution. Ocean current data resolution is a limitation, but the A. palmata test populations were distant enough to clearly identify varied leve ls of larval conn ectivity among them. The deterministic nature of the D flow routing, and the TauDEM upstream dependence computation does not represent horiz ontal and vertical mixi ng of larvae over
108 time; it was assumed A. palmata larval transport occurs by simple advection. This limitation in simulating the flow of water a nd larvae is not perfect for representing the mixing, transport, and dispersal of larvae that occurs naturall y in the ocean. However, the D results provided average snap shots or observations of dominant flows in the study area well enough for the analysis objectives to be successfully completed. There were also limitations on empirical data availabil ity and knowledge of the early life history of corals, including A. palmata The true maximum and minimum larval transport distance of A. palmata in the study area is unknown. The results heavily relied on a study by Shanks et al. (2003) which concluded larvae of mari ne invertebrates with long-di stance dispersal strategies tend to settle within 25 km of their spawning location. The results were reasoned to be applicable to any long-distance disper sing marine invertebrate, including A. palmata and many other broadcasting stony coral species which make-up the f oundation of the coral habitats within the study area. Another limita tion was the availability of empirical data on the spawning potential of the 34 validated A. palmata populations within the study area. However, the equal weighting of these populations when considering larval connectivity was acceptable and provided sign ificant findings that levels of larval connectivity among each A. palmata population at Horseshoe, Little Grecian, and Sand Island Reefs and the other 31 validated A. palmata populations significantly differ. Usefulness of this Research This thesis is an example of how GIS c ontinues to evolve in its modeling tools. The models of larval transport and larval connectivity presente d for the first time in this
109 study can be distributed and custom tailored fo r various input data fr om any region in the world. This will be a powerful tool for pol icy makers and environmental managers with various goals to understand f actors that influence larval connectivity among critical populations for the ultimate de sign of protected area networks. For example, managers and stakeholders can use the results from the present simulations to assist with identifying and delineating new MPAs in th e FKNMS. In addition, marine resource managers can use the A. palmata larval connectivity simulation results to modify their A. palmata restoration efforts to help enhan ce existing downstream populations. With additional GIS application innovations and more empirical data, this model will be further refined and validated, and it too will evolve over time. The conclusions further our understanding of the effects of biophysical processes on geographic patterns influenci ng population structure. Specif ically, it is demonstrated simulated larval connectivity may drive geographic patterns of clonal population structure in A. palmata Baums et al. (2006) indicate that it is not only the influx of larval recruits via larval connectivity, but the successf ul settlement, growth, and survival of new individuals are also influenced by biophysic al processes and dictate clonal population structure. However, for the first time, varia tions in simulated larval connectivity in the present study illustrate the initial step (i.e ., larval transfer) may influence geographic patterns of clonal popu lation structure in A. palmata These variations in simulated larval connectivity demonstrate the Horseshoe and Little Grecian Reef A. palmata populations are much less likely to recover from a dist urbance due to low c onnectivity with other upstream A. palmata populations, unlike the Sand Island Reef population.
110 All of these conclusions contribute to the understanding of th e interdependence and connectivity of coral habitats, A. palmata populations, and MPAs in the Florida Keys. The Florida Keys are a unique region of the world, where humans are highly dependent upon the KeyÂ’s ecosystems, in par ticular coral reefs. Ecosystems of the Florida Keys are in great decline, and if humans do not intervene and attempt to understand and protect these ecosystems, humank ind may lose them forever. Declining coral habitats and the effects on fisherie s and recreational oppor tunities (e.g., SCUBA diving) have a substantial socio-economic impact on the region. This thesis can help us in our efforts to preserve the interdependence and spatial connectedness of the regionÂ’s marine habitats through enhanced MPA design efforts, taking us one step closer to managing and protecting these environmental resources. Recommendations for Future Research The biophysical processes simulated (i.e., larval transport and larval connectivity) are dependent upon ocean current data resoluti on, scale, and accuracy. Results can be dramatically affected by the resolution of the ocean current input data. Therefore, research on the effects of higher resolution (s patially and temporally) ocean current data on the representation of larval connectivity is suggested. Map scale can also affect results. The study area and focal speciesÂ’ larval transport distance generally determine scale. For example, a broadcasting coral species would be studied on a much smaller scale than a brooding coral specie s. A smaller scale study would allow flows from greater distance to influence the resu lts, where a larger scale study would potentially ne glect contributing areas. If studying connectivity among
111 both long and short distance larval dispersers a compromise on scale may need to be made, so determining the effects of scale on th e results would be impor tant. The scale of the study should be based on the larval trans port strategy of the sp ecies or group of study organisms. Because it is possible A. palmata larvae may disperse further than 25 km, a smaller scale study analyzing flows from cora l greater than 25 km is recommended to determine the effects on larval connectivity with coral habitats even further upstream than those considered in the present study. Further SoFLA-HYCOM validation and p eer reviewed publications of SoFLAHYCOM simulations may also help improve la rval connectivity results. Using further validated and peer reviewed SoFLA-HYCOM simulations as input data would instill confidence in the accuracy of the larval connectivity analysis results. Continued larval connectivity model va lidation with additional species is recommended. For example, the same analys is could be conducted using the brooding short-distance dispersing coral Agaricia agaricites Validated A. agaricites populations and empirical genetic data would allow for additional validation of whether simulated larval connectivity also positive relationship with A. agaricites clonal diversity. In addition, this would allow for measuring the e ffects of scale since the analysis would be on a larger scale because A. agaricites is a short-distance dispersing species. Further validation using validated Acropora cervicornis populations and empirical genetic data is also recommended. Acropora cervicornis is closely related to A. palmata and it would be an excellent test to determine if larval c onnectivity also varies wi th clonal diversity in this species of coral.
112 The deterministic nature of D flow routing and the TauDEM upstream dependence computation are limited in repres enting biophysical behavior present in the natural environment. Further research into how horizontal an d vertical mixing of larvae over time is recommended. Additional research into scripting the D flow routing algorithm to model circulating flows would be beneficial. These circulating flows could then be input for the upstream dependence func tion which can represen t mixing of larvae. Methods for incorporating time and compu ting the degradation and accumulation of larvae over distance and time are also neede d. The ability to compute accumulating and dissipating larval flow (e.g., eith er all or a fraction of flow from cell A can go to cell B, then to cell C, and then back to cell A) ove r time would better repres ent ocean circulation and larval dispersal patterns resulting in improved larval connectivity estimations. Finally, further empirical data on larval biology/ecology/behavi or and the genetic connectedness of distant populations of any marine species are badly needed. A good place to start is to accumulate additional genetic data on A. palmata populations throughout the study area, and then attempt furt her validation of the larval connectivity model presented in this thesis. It would very interesting to further validate and determine how much larval connectivity influences the genetic structure of distant populations.
113 Literature Cited Airam, S., Dugan, J.E., Lafferty, K.D., Les lie, H., McArdle, D.A., & Warner, R.R. (2003). Applying ecological cr iteria to marine reserve design: A case study from the California Channel Islands. Ecological Applications, 13 (1), 170-184. ArcGIS 9.1. Environmental Systems Resear ch Institute (ESRI). Redlands, CA. Baker, J., Shepherd, S., & Edyvane, K. (1996) The use of marine fishery reserves to manage benthic fisheries, with emphasis on the South Australian abalone fishery. In R. Thackway (Ed.), Developing AustraliaÂ’s representative system of marine protected areas: Criteria and guideli nes for identificat ion and selection (pp. 103113). Canberra, Australia: Australi an Nature Conservation Agency. Bange, D.W., & Hoefer, J.N. (1976). A measur e of connectivity for geographic regions. The Professional Geographer, 28 (4), 362-370. Baums, I.B., Hughes, C.R., & Hellberg, M.E. (2005a). Mendelian micr osatellite loci for the Caribbean coral Acropora palmata Marine Ecology Progress Series, 288 115-127. Baums, I.B., Miller, M.W., & Hellberg, M.E. (2005b). Regionally isolated populations of an imperiled Caribbean coral, Acropora palmata Molecular Ecology, 14 1377-1390. Baums, I.B., Miller, M.W., & Hellberg, M.E. (2006). Geographic variation in clonal structure in a reef bu ilding Caribbean coral, Acropora palmata Ecological Monographs, 76 (4), 503-519. Becker, B.J., Fodrie, F.J., McMillan, P.A., & Levin, L.A. (2005). Spatial and temporal variation in trace elemental fingerprints of mytilid mussel shells: A precursor to invertebrate larval tracking. Limnology and Oceanography, 50 (1), 48-61. Bell, D. (2005). The value and importance of Geography. Primary Geographer, 56 4. Bohnsack, J.A. (1993). Marine reserves: Th ey enhance fisheries, reduce conflicts and protect resources. Oceanus, 36 (3), 63-71.
114 Botsford, L.W., Micheli, F., & Hastings, A. (2003). Principles for the design of marine reserves. Ecological Applications, 13 (1), 25-31. Brazeau, D.A., Sammarco, P.W., & Gleason, D.F. (2005). A multi-locus genetic assignment technique to assess sources of Agaricia agaricites larvae on coral reefs. Marine Biology, 147 (5), 1141-1148. Bunn, A.G., Urban, D.L., & Keitt, T.H. ( 2000). Landscape connec tivity: A conservation application of graph theory. Journal of Environmental Management, 59 265-278. Cantwell, M.D., & Forman, R.T.T. (1993). Landscape graphs: Ecolog ical modeling with graph theory to detect configurations common to diverse landscapes. Landscape Ecology, 8 (4), 239-255. Colin, P.L., Sadovy, Y.J., & Domeier, M.L. (2003). Manual for the study and conservation of reef fish spawning aggregations (Version 1). Fallbrook, CA: Society for the Conservation of Reef Fish Aggregations. Cook, R.R., & Auster, P.J. (2005). Use of si mulated annealing for identifying essential fish habitat in a multispecies context. Conservation Biology, 19 (3), 876-886. Cowen, R.K., Lwiza, K.M.M., Sponaugle, S ., Paris, C.B., & Olson, D.B. (2000). Connectivity of marine populations: Open or closed? Science, 287 857-859. Cowie-Haskell, B.D., & Delaney, J.M. (2003). Integrating science into the design of the Tortugas Ecological Reserve. Marine Technology Society Journal, 37 (1), 68-79. Criales, M.M., Wang, J., Browder, J.A., Jackso n, T., Robblee, M., & Hittle, C. (2003). Postlarval transport of pink shrimp into Florida Bay. Florida Bay Program & Abstracts: Joint Conf. on the Science and Restoration of the Greater Everglades and Florida Bay Ecosystem Â“From Kissimmee to the KeysÂ” (pp. 13-18). Domeier, M.L. (2004). A potential larval recruitment pathway originating from a Florida marine protected area. Fisheries Oceanography, 13 (5), 287-294. Dugan, J., & Davis, G. (1993). Applications of marine refugia to coastal fisheries management. Canadian Journal of Fisheries and Aquatic Sciences, 50 20291039. Eastman, R. (2001). The evoluti on of modeling tools in GIS. Directions Magazine, July, 18, 2001 Retrieved January 22, 2007, from Directions Magazine Web site: http://www.directionsmag.com/a rticle.php?article_id=114&trv=1 Fadallah, Y.H. (1983). Sexual reproduc tion, development and larval biology in scleractinian corals: A review. Coral Reefs, 2 129-150.
115 Federal Register. (2006). Endangered and threatened species: Final listing determinations for Elkhorn Coral and Staghorn Coral. Federal Register, 71 (89), 26852-26872. Federal Register. (2000). Executive Or der 13158 of May 26, 2000: Marine protected areas. Federal Register, 65 (105), 34909-34911. Florida Fish and Wildlife Cons ervation Commission, Fish and Wildlife Research Institute (FWRI). (1998). Benthic Habitats of the Florida Keys St. Petersburg, FL: Coastal and Marine Resource Assessment Florida Fish and Wildlife Research Institute. Florida Keys National Marine Sanctuary (F KNMS). (2005). Retrieved on December 28, 2005, from The Florida Keys National Marine Sanctuary Web site: http://www.fknms.nos.noaa.gov/ Franklin, C.F. (2002). Br ing your GIS: Lessons learne d from planning a marine protected area. Presented at the 5th Annual SCGIS Conference, July 5-7, 2002. The Society for Conservation GIS (S CGIS) 2002 Annual Conference Abstracts Franklin, C.F., Ault, J.S., Smith, S.G., Luo, J., Meester, G.A., Diaz, G.A., Chiappone, M., Swanson, D.W., Miller, S.L., & Bohnsack, J.A. (2003). Benthic habitat mapping in the Tortugas region, Florida specia l issue on marine and coastal GIS. Marine Geodesy, 26 (1-2), 19-34. Gaines, S.D. (2005). Sensitivity to environmental di sturbance: Effects of larval dispersal Retrieved on November 20, 2006, from University of California at Santa Barbara, Marine Science Institute Web site: http://www.msi.ucsb.edu/msilinks /CRC/CRCtexts/Toxics/ucsb1.html Gerber, L.R., Botsford, L.W., Hastings, A ., Possingham, H.P., Gaines, S.D., Palumbi, S.R., & Andelman, S. (2003). Populati on models for marine reserve design: A retrospective and pros pective synthesis. Ecological Applications, 13 (1), S47-S64. Groves, C.R. (2003). Drafting a conserva tion blueprint: A practitionerÂ’s guide to planning for biodiversity. Wash ington D.C.: Island Press. Halpern, B.S. (2003). The impact of mari ne reserves: Do reserves work and does reserve size matter. Ecological Applications, 13 (1), S117-S137. Halpin, P.N., & Bunn, A.G. (2000). Using GI S to compute a least-cost distance matrix: A comparison of terrestrial and ma rine ecological applications. Twentieth Annual ESRI User Conference. Retrieved on January 3, 2006, from ESRI Web site: http://gis.esri.com/library/userconf/p roc00/professional/papers/PAP890/p890.htm
116 Harary, F. (1969). Graph Theory. Read ing, MA: Addison-Wesley Publishing Company. Hoegh-Guldberg, O. (1999). Climate change, coral bleaching and the future of the worldÂ’s coral reefs. Marine & Freshwater Research, 40 839-866. Jackson, P., & Massey, D. (2006) Thinking geographically. Geography, 91 (3), 199204. Jones, G.P., Milicich, M.J., Emslie, M.J., & Lunow, C. (1999). Self-recruitment in a coral reef fish population. Nature, 402 802-804. Jones, D.L., Lara, M.R., & Lamkin, J.T. (2003). Use of geochemical tracers to elucidate life history trajectories of gray snapper within south FloridaÂ’ s marine ecosystems Joint conference on the scie nce and restoration of th e greater Everglades and Florida Bay ecosystem, Florida Bay and Adjacent Marine Ecosystems Program Management Committee, Palm Harbor, FL. Jones, D.L., Lara, M.R., Yeung, C., Criales, MM., Jackson, T.L., & Richards, W.J. (2005). Coastal processes affecting larval supply into Florida Bay: Snappers (Pisces: Lutjanidae) Retrieved on January 3, 2006, from National Oceanic and Atmospheric Administration, Southeast Fisheries Science Center Web site: http://www.sefsc.noaa.gov/PDFdocs/Jones_etal_2001.pdf Kansky, K. (1963). The structur e of transportation networks. University of Chicago, Department of Geography, Research Paper, 84. Kendall Jr., A.W., & Picquelle, S. J. (2003). Marine protect ed areas and the early life history of fishes. Alaska Fisheries Science Ce nter Processed Report 2003-10 Knowlton, N. (2001). The future of coral reefs. Proceedings of the National Academy of Sciences of the Unit ed States of America, 98 (10), 5419-5425. Kourafalou, V.H., Balotro, R.S., & Lee, T.N. (2005). The SoFLA-HYCOM (South Florida HYCOM) Regional Model around the St raits of Florida, Florida Bay and the Florida Keys. UM/RSMAS Technical Report, 2005-03 28. Lee, T.N., Clarke, M.E., Williams, E., Szmant, A.F., & Berger, T. (1994). Evolution of the Tortugas Gyre and its influence on the recruitment in the Florida Keys. Bulletin of Marine Science, 54 (3), 621-646. Lee, T.N., Johns, E., Wilson, D., & Williams, E. (1999). Site characterization for the Tortugas region: Physical oceanography and recruitment. Report Miami, FL: University of Miami, Rosenstiel School of Marine Science.
117 Lee, T.N., & Williams, E. (1999). Mean dist ribution and seasonal variability of coastal currents and temperature in the Florida Keys with implications for larval recruitment. Bulletin of Marine Science, 64 (1), 35-56. Lee, T.N., Williams, E., Johns, E., Wilson, D., & Smith, N.P. (2002). Transport processes linking south Florida coastal ecosys tems. In J.W. Porter, & K.G. Porter (Eds.), The Everglades, Florida Bay and cora l reefs of the Florida Keys: An ecosystem source book (pp. 309-342). Boca Raton, FL: CRC Press. Leslie, H., Ruckelshaus, M., Ball, I.R., Andelman, S., & Possingham, H.P. (2003). Using siting algorithms in the desi gn of marine reserve networks. Ecological Applications, 13 (1), S185-S198. Leslie, H.M. (2005). A synthesis of ma rine conservation planning approaches. Conservation Biology, 19 (6), 1701. Lindeman, K.C., Pugliese, R., Waugh, G.T., & Ault, J.S. (2000). Developmental patterns within a mutispecies reef fishery: Management applications for essential fish habitats and protected areas. Bulletin of Mari ne Science, 66 (3), 929-956. Lubchenco, J., Palumbi, S.R., Gaines, S.D., & Andelman, S. (2003). Plugging the hole in the ocean: The emerging science of marine reserves. Ecological Applications, 13 S3-S7. Matisziw, T.C., & Murray, A.T. (2006). Pr omoting species persistence through spatial association optimization in nature reserve design. Journal of Geographical Systems, 8 (3), 289-305. Meester, G.A., Mehrotra, A., Ault, J.S., & Baker, E.K. (2004). Designing marine reserves for fishery management. Management Science, 50 (8), 1031-1043. Michels, E., Cottenie, K., Neys, L., De Gelas, K., Coppin, P., & De Meester, L. (2001). Geographical and genetic distances am ong zooplankton populations in a set of interconnected ponds: A plea for usin g GIS modelling of the effective geographical distance. Molecular Ecology, 10 (8), 1929-1938. Miller, S.L., Chiappone, M., Rutten, L.M., & Swanson, D.W. (in preparation). Population assessment of staghorn (Acropoa cervicornis) and elkhorn corals (A. palmata) in the upper Keys region of the Florida Keys National Marine Sanctuary Wilmington, NC: Center for Mari ne Science, University of North Carolina at Wilmington. Moilanen, A., & Nieminen, M. (2002). Simple connectivity measures in spatial ecology. Ecology, 83 (4), 1131-1145.
118 Moore, D.S. (1986). Tests of chi-squared type. In M.A. Stevens, & R.B. DÂ’Agnostino (Eds.), Goodness-of-fit techniques New York, New Yo rk: Marcel Dekker. Mora, C., & Sale, P.F. (2002). Are populati ons of coral reef fi sh open or closed? Trends in Ecology and Evolution, 17 (9), 422-428. Morgan, L., Etnoyer, P., Wilkinson, T., Herrman n, H., Tsao, F., & Maxwell, S. (2003). Identifying priority conservation areas from Baja California to the Bering Sea. In N.W.P. Munro, P. Deardon, T.B. Herm an, K. Beazley, & S. Bondrup-Nielsen (Eds.), Proceedings of the Fifth Inter national Conference on Science and Management of Protected Areas May Wolfville, Nova Scotia, Canada: Science and Management of Protect ed Areas Association. Mullineaux, L.S., DiBacco, C., Lerczak, J., T horrold, S.R., Neubert, M., Caswell, H., Levin, L.A., & Largier, J.L. (in preparation). Connectivity in bivalve populations: Assessing sources of larval recruits. Retrieved on January 3, 2006, from Woods Hole Oceanographic Institution Web site: http://www.whoi.edu/science/B/pe ople/lmullineaux/Biv_propsum.pdf Mumby, P.J., & Dytham, C. (2006). Metapopulation dynamics of hard corals. In J.P. Kritzer, & C. Dytham (Eds.), Marine Metapopulations (pp. 157-196). Burlington, MA: Elsevier Academic Press. National Marine Protected Areas Center (NMPAC). (2004). An inventory of GIS-based decision-support tools for MPAs Retreived on January 3, 2006, from National Oceanic and Atmospheric Administration, Coastal Services Center Web site: http://www.csc.noaa.gov/mpa/ FINAL_DecisionSupRpt.pdf National Marine Sanctuary Program (NMSP). (2005). About your National Marine Sanctuaries Retrieved on December 3, 2005 fr om National Marine Sanctuary Program Web site: http:// www.sanctuaries.nos.noaa.gov National Oceanic and Atmospheric Administ ration (NOAA), Centers for Coastal Ocean Science, Center for Coastal Monito ring and Assessment (CCMA) Biogeography Team. Silver Spring, MD. NO AA Biogeography Team Web site: http://ccmaserver.nos.noaa.gov/a bout/biogeography/welcome.html National Oceanic and Atmospheric Admini stration (NOAA), National Ocean Service (NOS). Silver Spring, MD. NOAA NOS Web site: http://www.oceanservice.no aa.gov/about/welcome.html National Research Council. (2001). Marine protected areas: Tools for sustaining ocean ecosystems Washington, D.C.: The National Academies Press. Palumbi, S.R. (1999). The prodigal fish. Nature, 402 733-735.
119 Palumbi, S.R., Gaines, S.D., Leslie, H., & Warner, R.R. (2003). New wave: High-tech tools to help marine reserve research. Frontiers in Ecology and the Environment, 1 (2), 73-79. Roberts, C., & Polunin, N.V.C. (1992). E ffects of marine fisher y reserve protection on northern Red Sea fish populations. Proceedings of the Seventh International Coral Reef Symposium, 2 969-977. Roberts, C.M. (1997). Connectivity and management of Caribbean coral reefs. Science, 278 1454-1457. Roberts, C.M., Bohnsack, J.A., Gell, F., Hawkin s, J.P., & Goodridge, R. (2001). Effects of marine reserves on adjacent fisheries. Science, 294 1920-1923. Sale, F. (2006). Connectivity Retrieved on December 28, 2006, from Coral Reef Targeted Research & Capacity Building Program Web site: http://www.inweh.unu.edu/inweh/Coastal/Co ralReef/Connectivity_Brochure.pdf Salm, R.V., Clark, J., & Siirila, E. (2000). Marine and coastal protected areas: A guide for planners and managers (p. 371). Washington, DC: International Union for the Conservation of Nature and Natural Resources. Sammarco, P.W., Atchison, A.D., & Boland, G.S. (2004). Expansion of coral communities within the Northern Gulf of Mexico via offshore oil and gas platforms. Marine Ecology Progress Series, 280 129-143. Shanks, A.L., Grantham, B.A., & Carr, M.H. (2003). Propagule di spersal distance and the size and spacing of marine reserves. Ecological Applications, 13 (1), S159S169. Siegel, S., & Castellan, N.J. (1988). Nonparametric statistics fo r the behavioral sciences (2nd ed.) (pp. 174-183). New York, NY: McGraw Hill. Stobutzki, I.C. (2000). Mari ne reserves and the complexity of larval dispersal. Reviews in Fish Biology and Fisheries, 10 (4), 515-518. Swearer, S.E., Caselle, J.E., Lea, D.W., & Wa rner, R.R. (1999). Larval retention and recruitment in an island populat ion of a coral-reef fish. Nature, 402 799-802. Tarboton, D.G. (1997). A new method for th e determination of flow directions and contributing areas in grid digital elevation models. Water Resources Research, 33 (2), 309-313.
120 Tarboton, D.G. (2005). Terrain analysis using digita l elevation models (TauDEM) Retrieved on March 1, 2006, from Utah St ate University, Utah Water Research Laboratory Web site: http:// hydrology.neng.usu.edu/taudem/ Thibaut, E., Lagadeuc, Y., Olivier, F., Da uvin, J.C., & Retire, C. (1998). Do hydrodynamic factors affect the recruitm ent of marine invertebrates in a macrotidal area? Hydrobiologia, 375 165-176. Tischendorf, L., & Fahrig, L. (2000). On the usage and measurement of landscape connectivity. Oikos, 90 (1), 7-19. Todd, C.D. (1998). Larval supply and recruitm ent of benthic invertebrates: Do larvae always disperse as much as we believe? Hydrobiologia, 375 1-21. Urban, D.L. (2000). Using model analysis to design monitoring programs for landscape management and impact assessment. Ecological Applications, 10 (6), 1820-1832. Urban, D., & Keitt, T. (2001). Landscape connectivity: A graph th eory perspective. Ecology, 82 (5), 1205-1218. U.S. Department of Comme rce (USDOC). (1996). Strategy for stewardship: Florida Keys National Marine Sanctuary final management plan/e nvironmental impact statement, Volume II. Retrieved on November 20, 2006, from National Oceanic and Atmospheric Administration, National Ocean Service Web site: http://www.fknms.nos.noaa.gov/regs/fmp2.pdf U.S. Department of Comme rce (USDOC). (2000). Strategy for stewardship: Tortugas Ecological Reserve final supplemental environmental impac t statement/final supplemental management plan. Retrieved on November 20, 2006, from National Oceanic and Atmospheric Administrati on, National Ocean Service Web site: http://www.fknms.nos.noaa.gov/regs/FinalFSEIS.pdf Warner, R.R., & Cowen, R.K. (2002). Lo cal retention of pr oduction in marine populations: Evidence, mechanisms, and consequences. Bulletin of Marine Science, 70 (1), 245-249. Wells, F., & Keesing, J. (1990). Popul ation characteristics of the abalone Haliotis roei on intertidal platforms in th e Perth metropolitan area. J ournal of the Malacological Society of Australia, 11 65-71. Yeung, C., Jones, D., Criales, M., Lara, M., Lamkin, J., Richards, W., & Jackson, T. (2005). Large eddy, small eddy Â– supplying recruits to the south Florida ecosystem. Poster retrieved on November 20, 2006, from National Oceanic and Atmospheric Administration, Southeast Fisheries Science Center Web site: http://www.sefsc.noaa.gov/PDFdocs/GCFI55.pdf
122 Appendix A Cross-validation of Interpolated Grids The SoFLA-HYCOM day 213 current vector points are shown in Figure A1. A spline (tension) interpolation grid of day 213 U values overlaid with 5% of the true points from day 213 is shown in Figure A2. The day 213 values were also used to compute Krigin and IDW (both with ArcGIS Spatial An alyst default settings ) interpolation grids for cross-validation using 5% of the day 213 points for comparisons among the interpolated and true U values. The mean of the (true Â– inte rpolated U component) values are slightly overestimated by the IDW and Krigin techniques, and slightly underestimated by the Spline technique (Table A1). Based on this measure, the Krigin interpolation technique is clos est to zero, which means it may be best at estimating the true U component value.
123 Appendix A (Continued) Figure A1. SoFLA-HYCOM Curre nt Vector Point Features
124 Appendix A (Continued) Figure A2. SoFLA-HYCOM U Current Vect or Component Spline Interpolation and Point Features used for Cross-Validation 24.22 22.33 18.776 70.435 -3.0527 -2.4915 27.8689 43.0199 -2.4017 -1.4536 44.2547 46.0587 -2.4416 -3.6226 24.3135 52.9864 -3.3102 53.1276 26.7085 59.7614 30.9499 63.7649 65.0571 56.1603 67.4452 39.4478 Spline Tension (U Component) -6.391319275 3.4783766 3.478376601 13.34807248 13.34807249 23.21776835 23.21776836 33.08746423 33.08746424 42.9571601 42.95716011 52.82685598 52.82685599 62.69655185 62.69655186 72.56624773 72.56624774 82.4359436 5% of SoFLA-HYCOM Day 001 Points 5% of the Day 213 Points
125 Appendix A (Continued) Table A1. Cross-Validation of Interpolated Grids IDW Krigin Spline True Interpolated U Component Minimum -4.30 -2.92 -3.25 Maximum 5.79 5.58 5.88 Mean 0.30 0.03 -0.19 Standard Deviation 2.18 1.76 1.81 Abs (True Interpolated U Component) Minimum 0.01 0.01 0.00 Maximum 5.79 5.58 5.88 Mean 1.51 1.19 1.13 Standard Deviation 1.57 1.27 1.40 The standard deviations of the (true Â– interpolated U component) values are similar among the three techniques. Based on this measure, the Krigin interpolation technique, with a value of 1.76 is slightly better th an the other techniques (Table A1). The mean of the absolute value of the (t rue Â– interpolated U component) values is the best indicator of the error in each of the inte rpolation grids. Based on this measure, the Spline (tension) technique with a value of 1.13 was sligh tly better at estimating the true U component values in this case (Table A1).
126 Appendix B Figure B1. Larval Transport and Connectiv ity Analytic Model: Analyses One and Two
127 Appendix B (Continued) Figure B2. Larval Transport a nd Connectivity Analytic Model: An alyses One and Two (Continued)
128 Appendix B (Continued) Figure B3. Larval Transport a nd Connectivity Analytic Model: An alyses One and Two (Continued)
129 Appendix B (Continued) Figure B4. Larval Transport a nd Connectivity Analytic Model: An alyses One and Two (Continued)
130 Appendix C Table C1. Daily Contributing Flow Valu es from All Coral Habitats to Each A. palmata Test Population A. palmata Test Populations August Day Sand Island Little Grecian Horseshoe 1 0.015920 0.002596 0.001141 2 0.003912 0.001714 0.002030 3 0.005329 0.008294 0.001297 4 0.009405 0.003795 0.002082 5 0.030878 0.005894 0.001371 6 0.035923 0.001638 0.000746 7 0.025723 0.003980 0.000851 8 0.055664 0.006700 0.001433 9 0.029025 0.011972 0.001459 10 0.034164 0.008229 0.004007 11 0.029389 0.004773 0.001051 12 0.037068 0.002102 0.000503 13 0.042049 0.003812 0.000967 14 0.039840 0.002880 0.001033 15 0.047842 0.002387 0.000681 16 0.054213 0.001704 0.000411 17 0.045786 0.002820 0.001016 18 0.037309 0.003028 0.001380 19 0.010412 0.005345 0.003873 20 0.010505 0.016040 0.007343 21 0.018521 0.005815 0.003953 22 0.022788 0.005610 0.003978 23 0.028987 0.005064 0.003284 24 0.023210 0.005290 0.003306 25 0.012149 0.007759 0.004711 26 0.009630 0.018207 0.007915 27 0.010297 0.009638 0.004699 28 0.010104 0.010810 0.005429 29 0.012016 0.010406 0.005689 30 0.016005 0.009719 0.005717 31 0.025862 0.007746 0.005384
131 Appendix D Figure D1. Description of B ox-Plots (from Analyse-It for Microsoft Excel Help Index)
132 Appendix E Table E1. Daily Contributing Flow Values from Validated A. palmata Populations to Each A. palmata Test Population A. palmata Test Populations August Day Sand Island Little Grecian Horseshoe 1 0.116338 0.033218 0.031532 2 0.142142 0.027273 0.049601 3 0.089912 0.027812 0.036589 4 0.134904 0.032275 0.031532 5 0.115024 0.051095 0.031532 6 0.117984 0.047971 0.031532 7 0.122196 0.043961 0.031532 8 0.122541 0.033122 0.031532 9 0.109649 0.039906 0.031532 10 0.116796 0.028053 0.031532 11 0.118087 0.046293 0.031532 12 0.115896 0.052962 0.031532 13 0.116461 0.052915 0.031532 14 0.116956 0.051836 0.031532 15 0.119762 0.056930 0.031532 16 0.128938 0.056541 0.031532 17 0.127331 0.048744 0.031532 18 0.122127 0.046909 0.031532 19 0.109902 0.027277 0.033324 20 0.111050 0.030741 0.035156 21 0.108029 0.027897 0.031577 22 0.112346 0.027273 0.031532 23 0.118352 0.027409 0.031532 24 0.112502 0.028198 0.031532 25 0.105081 0.027339 0.031532 26 0.111313 0.031220 0.033899 27 0.109622 0.027320 0.031805 28 0.106427 0.027693 0.032430 29 0.104652 0.027540 0.033570 30 0.105748 0.027280 0.035538 31 0.113770 0.027273 0.040201
133 Appendix F Table F1. Daily Contributing Flow Values from Each Coral Habitat Type to MPAs. (Coral habitat types are A) Patch Reefs A ggregated, B) Patch Reefs Aggregated with Halo, C) Patch Reefs Â– Coral or Rock Patche s with Bare Sand, D) Patch Reefs Halo, E) Patch Reefs Individual, F) Platform Margin Reefs Â– Back Reef, G) Platform Margin Reefs Â– Drowned Spur and Groove, H) Plat form Margin Reefs Â– Reef Rubble, I) Platform Margin Reefs Â– Remnant Â– Low Profile, and J) Platform Margin Reefs Â– Shallow Spur and Groove.) Contributing Flow to MPAs by Habitat Type Day A B C D E F G H I J 1 0.1581 0.1581 0.0289 0.0126 0.1725 0.0428 0.0045 0.0134 0.0506 0.0530 2 0.0644 0.0644 0.0309 0.0107 0.0712 0.0000 0.0000 0.0000 0.0466 0.0001 3 0.0483 0.0483 0.0272 0.0001 0.0812 0.0005 0.0073 0.0008 0.0108 0.0004 4 0.1371 0.1371 0.1230 0.0058 0.2040 0.0261 0.0374 0.0314 0.1119 0.0285 5 0.0867 0.0867 0.1053 0.0949 0.1990 0.0920 0.0505 0.0315 0.0768 0.2068 6 0.1430 0.1430 0.2172 0.0639 0.1835 0.1555 0.1168 0.1176 0.2124 0.2444 7 0.1400 0.1400 0.2867 0.2010 0.1521 0.1553 0.1882 0.1574 0.2692 0.2488 8 0.1021 0.1021 0.4124 0.0441 0.1408 0.1609 0.3269 0.2354 0.4156 0.2913 9 0.2195 0.2195 0.3655 0.0220 0.1950 0.0049 0.2580 0.1591 0.3111 0.0286 10 0.2506 0.2506 0.4699 0.1630 0.3233 0.1483 0.3656 0.2475 0.4191 0.2461 11 0.1010 0.1010 0.3851 0.0991 0.2639 0.1665 0.3639 0.2526 0.3148 0.2882 12 0.0160 0.0160 0.3447 0.1354 0.2336 0.1594 0.2575 0.2100 0.2988 0.2901 13 0.0445 0.0445 0.3963 0.1425 0.2689 0.1516 0.2960 0.2119 0.3682 0.2808 14 0.0276 0.0276 0.3615 0.1000 0.1804 0.1560 0.2461 0.1812 0.3561 0.2863 15 0.0373 0.0373 0.3544 0.0974 0.2161 0.1642 0.2297 0.1831 0.3826 0.2937 16 0.0479 0.0479 0.2341 0.0273 0.1534 0.1631 0.2135 0.2039 0.3191 0.2671 17 0.0657 0.0657 0.4001 0.1227 0.3100 0.1517 0.3257 0.2378 0.4397 0.2404 18 0.0367 0.0367 0.2312 0.0209 0.0848 0.1711 0.2506 0.1465 0.1403 0.2987 19 0.1031 0.1031 0.3785 0.0279 0.1468 0.0019 0.2495 0.1144 0.2479 0.0114 20 0.0928 0.0928 0.3663 0.0272 0.1816 0.1148 0.2923 0.1470 0.2592 0.1868 21 0.2565 0.2565 0.5229 0.1055 0.3179 0.0771 0.3763 0.1961 0.4730 0.1899 22 0.1725 0.1725 0.3942 0.0700 0.2015 0.1262 0.2988 0.1321 0.3393 0.2567 23 0.1809 0.1809 0.3977 0.0949 0.2028 0.1566 0.3571 0.2420 0.3638 0.2885 24 0.1824 0.1824 0.4262 0.0925 0.1815 0.1279 0.3409 0.2178 0.3803 0.2583 25 0.2547 0.2547 0.4759 0.0776 0.3067 0.0202 0.3471 0.1788 0.3889 0.0708 26 0.1673 0.1673 0.4511 0.0969 0.3252 0.1302 0.3687 0.2189 0.3904 0.2051 27 0.1997 0.1997 0.5733 0.1491 0.4153 0.0193 0.4218 0.2475 0.4967 0.0302 28 0.1664 0.1664 0.5258 0.0608 0.3059 0.0231 0.3987 0.2413 0.4564 0.0260 29 0.0933 0.0933 0.4180 0.0695 0.3128 0.0316 0.3321 0.1909 0.3163 0.0607 30 0.0959 0.0959 0.5449 0.1777 0.3358 0.0633 0.3945 0.2115 0.4390 0.1485 31 0.0739 0.0739 0.4116 0.1542 0.2937 0.1371 0.3481 0.2049 0.3128 0.2684
134 Appendix G Table G1. Mann-Whitney All-Pairwise Compar ison Test for Differences in Coral Larval Connectivity with MPAs among Coral Habitat Types Mann-Whitney test (normal approximations, and corrected for ties) 2-tailed p Patch Reefs Aggregated v Patch Re efs Aggregated with Halo 1.0000 Patch Reefs Aggregated v Patch Reefs Cora l or Rock Patches with Bare Sand <0.0001 Patch Reefs Aggregated v Patch Reefs Halo 0.0307 Patch Reefs Aggregated v Pa tch Reefs Individual <0.0001 Patch Reefs Aggregated v Platform Margin Reefs Back Reef 0.2128 Patch Reefs Aggregated v Platform Margin Reefs Drowned Spur and Groove <0.0001 Patch Reefs Aggregated v Platform Margin Reefs Reef Rubble 0.0177 Patch Reefs Aggregated v Platform Marg in Reefs Remnant Low Profile <0.0001 Patch Reefs Aggregated v Platform Marg in Reefs Shallow Spur and Groove 0.0247 Patch Reefs Aggregated with Ha lo v Patch Reefs Coral or Rock Patches with Bare Sand <0.0001 Patch Reefs Aggregated with Ha lo v Patch Reefs Halo 0.0307 Patch Reefs Aggregated with Halo v Patch Reefs Individual <0.0001 Patch Reefs Aggregated with Halo v Pl atform Margin Reefs Back Reef 0.2128 Patch Reefs Aggregated with Halo v Platform Margin Reefs Drowned Spur and Groove <0.0001 Patch Reefs Aggregated with Halo v Pl atform Margin Reefs Reef Rubble 0.0177 Patch Reefs Aggregated with Halo v Platform Margin Reefs Remnant Low Profile <0.0001 Patch Reefs Aggregated with Halo v Platform Margin Reefs Shallow Spur and Groove 0.0247 Patch Reefs Coral or Rock Patches with Bare Sand v Patch Reefs Halo <0.0001 Patch Reefs Coral or Rock Patches with Bare Sand v Patch Reefs Individual 0.0001 Patch Reefs Coral or Rock Patches with Bare Sa nd v Platform Margin Reefs Back Reef <0.0001 Patch Reefs Coral or Rock Patches w/ Bare Sand v Platf. Margin Reefs Drd Spur & Groove 0.0033 Patch Reefs Coral or Rock Patches w/ Bare Sa nd v Platf. Margin Reefs Reef Rubble <0.0001 Patch Reefs Coral or Rock Patches w/ Bare Sand v Platf. Margin Reefs Remnt Low Profile 0.1571 Patch Reefs Coral or Rock Patches w/ Bare Sand v Plat Margin Reefs Shal Spur & Groove <0.0001 Patch Reefs Halo v Patch Reefs Individual <0.0001 Patch Reefs Halo v Platform Margin Reefs Back Reef 0.2342 Patch Reefs Halo v Platform Margin Reefs Drowned Spur and Groove <0.0001 Patch Reefs Halo v Platform Margin Reefs Reef Rubble <0.0001 Patch Reefs Halo v Platform Margin Reefs Remnant Low Profile <0.0001 Patch Reefs Halo v Platform Margin Reefs Shallow Spur and Groove 0.0009 Patch Reefs Individual v Platform Margin Reefs Back Reef <0.0001 Patch Reefs Individual v Platform Margin Reefs Drowned Spur and Groove 0.0378 Patch Reefs Individual v Platform Margin Reefs Reef Rubble 0.0448 Patch Reefs Individual v Platform Margin Reefs Remnant Low Profile 0.0030 Patch Reefs Individual v Platform Margin Reefs Shallow Spur and Groove 0.1881 Platform Margin Reefs Back Reef v Platform Margin Reefs Drowned Spur and Groove <0.0001 Platform Margin Reefs Back Reef v Platform Margin Reefs Reef Rubble 0.0002 Platform Margin Reefs Back Reef v Platform Margin Reefs Remnant Low Profile <0.0001 Platform Margin Reefs Back Reef v Platform Margin Reefs Shallow Spur and Groove 0.0014 Platform Margin Reefs Drowned Spur and Groove v Platform Margin Reefs Reef Rubble <0.0001 Platform Margin Reefs Drowned Spur & Groove v Platf. Margin Reefs Remnt Low Profile 0.1101 Platform Margin Reefs Drowned Spur & Groove v Pl atf. Margin Reefs Shal Spur & Groove 0.0033 Platform Margin Reefs Reef Rubble v Platform Margin Reefs Remnant Low Profile <0.0001 Platform Margin Reefs Reef Rubble v Platform Margin Reefs Shallow Spur and Groove 0.1132 Platform Margin Reefs Remnant Low Profile v Plat f. Margin Reefs Shallow Spur & Groove <0.0001