Skip to main content
Internet Archive's 25th Anniversary Logo

Full text of "Characterization of the Benthos, Marine Debris and Bottom Fish at Gray's Reef National Marine Sanctuary"

See other formats


Characterization of the 

Benthos, Marine Debris and Bottom Fish at 

Gray's Reef National Marine Sanctuary 

by Matthew S. Kendall, Laurie J. Bauer and Christopher F.G. Jeffrey 
In partnership with Gray's Reef National Marine Sanctuary 







January 2007 



NOAA Technical Memorandum NOS NCCOS 50 



Mention of trade names or commercial products does not constitute endorsement or recommendation for their 
use by the United States Government. 



Citation: 

Kendall, M.S., L.J. Bauer and C.F.G. Jeffrey. 2007. Characterization of the Benthos, Marine Debris and Bottom 
Fish at Gray's Reef National Marine Sanctuary. Prepared by National Centers for Coastal Ocean Science (NC- 
COS) Biogeography Team in cooperation with the National Marine Sanctuary Program. Silver Spring, MD. NOAA 
Technical Memorandum NOS NCCOS 50. 82 pp. + Appendices. 



Characterization of the Benthos, Marine Debris and Bottom Fish at 
Gray's Reef National Marine Sanctuary 



Center for Coastal Monitoring and Assessment (CCMA) 

NOAA/NOS/NCCOS 

1305 East-West Highway, SSMC-4, N/SCI 1 

Silver Spring, MD 20910 

NOAA Technical Memorandum NOS NCCOS 50 

January 2007 

Authors: 

Matthew S. Kendall (NCCOS) 
Laurie J. Bauer (NCCOS) 
Christopher F.G. Jeffrey (NCCOS) 




United States Department National Oceanic and National Ocean Service 

of Commerce Atmospheric Administration 

Carlos M. Gutierrez Conrad C. Lautenbacher, Jr. John H. Dunnigan 

Secretary Administrator Administrator 



ABOUT THIS DOCUMENT 

This ecological characterization represents the continuation of an ongoing partnership between the National 
Marine Sanctuary Program (NMSP) and the National Centers for Coastal Ocean Science (NCCOS), Center for 
Coastal Monitoring and Assessment (CCMA). The purpose of this collaboration is to apply a biogeographical 
approach to the management of marine resources within the National Marine Sanctuaries. This particular work, 
conducted in consultation with Gray's Reef National Marine Sanctuary (GRNMS) and scientists conducting re- 
search within the South Atlantic Bight region, builds on and advances biogeographic techniques developed by 
CCMA's Biogeography Team for other National Marine Sanctuaries including Channel Islands, Cordell Bank, 
Gulf of Farallones, Monterey Bay, and Stellwagen Bank. At the onset of the project, CCMA, GRNMS, and NMSP 
staff identified a set of targeted research topics to fill existing gaps in baseline data, and enhance the understand- 
ing of key ecological patterns and processes to support the Sanctuary. 

The characterization consists of two complementary components: a text report that includes a suite of quantita- 
tive spatial and statistical analyses that characterize physical and biological features of GRNMS; and the raw da- 
tabase of all spatial data analyzed to conduct the characterization. The report provides essential information on 
the distribution of modeled and observed species and features needed to support the development of monitoring 
and scientific studies, the development of educational material, and support of other spatially-explicit manage- 
ment decisions. The results of this ecological characterization are available via website. For more information 
on this effort please visit the NCCOS Biogeography Team webpage dedicated to this project at http://ccma.nos. 
noaa.gov/ecosystems/sanctuaries/grays_nms.html or direct questions and comments to: 

Mark E. Monaco, Ph.D. 

Biogeography Team Lead 

National Oceanic and Atmospheric Administration 

1305 East West Highway 

SSMC 4, N/SCI-1 

Silver Spring, MD 20910 

Phone: (301) 713-3028x160 

Email: Mark.Monaco@noaa.gov 

Or 

Greg McFall 

Research Coordinator 

Gray's Reef National Marine Sanctuary 

National Oceanic and Atmospheric Administration 

10 Ocean Science Circle 

Savannah, GA 31411 

Phone:(912)598-2345 

Email: Greg.McFall@noaa.gov 




EXECUTIVE SUMMARY 



Baseline characterization of resources is an essential part of marine protected area (MPA) management and 
is critical to inform adaptive management. Gray's Reef National Marine Sanctuary (GRNMS) currently lacks 
adequate characterization of several key resources as identified in the 2006 Final Management Plan. The objec- 
tives of this characterization were to fulfill this need by characterizing the bottom fish, benthic features, marine 
debris, and the relationships among them for the different bottom types within the sanctuary: ledges, sparse live 
bottom, rippled sand, and flat sand. Particular attention was given to characterizing the different ledge types, 
their fish communities, and the marine debris associated with them given the importance of this bottom type to 
the sanctuary. 

The characterization has been divided into 
four sections. Section 1 provides a brief over- 
view of the project, its relevance to sanctuary 
needs, methods of site selection, and general 
field procedures. Section 2 provides the sur- 
vey methods, results, discussion, and recom- 
mendations for monitoring specific to the ben- 
thic characterization. Section 3 describes the 
characterization of marine debris. Section 4 is 
specific to the characterization of bottom fish. 
Field surveys were conducted during August 
2004, May 2005, and August 2005. A total of 
179 surveys were completed over ledge bot- 
tom (n=92), sparse live bottom (n=51), flat 
sand (n=20), and rippled sand (n=16). There 
were three components to each field survey: 
fish counting, benthic assessment, and quan- 
tification of marine debris. All components oc- 
curred within a 25 x 4 m belt transect. Two 
divers performed the transect at each survey 
site. One diver was responsible for identifica- 
tion of fish species, size, and abundance using a visual survey. The second diver was responsible for character- 
ization of benthic features using five randomly placed 1 m 2 quadrats, measuring ledge height and other benthic 
structures, and quantifying marine debris within the entire transect. 




Image 1a. Fish schooling around densely colonized live bottom in GRNMS. 



GRNMS is composed of four main bottom types: flat sand, rippled sand, sparsely colonized live bottom, and 
densely colonized live bottom (ledges). Independent evaluation of the thematic accuracy of the GRNMS benthic 
map produced by Kendall et al. (2005) revealed high overall accuracy (93%). Most discrepancies between map 
and diver classification occurred during August 2004 and likely can be attributed to several factors, including 
actual map or diver errors, and changes in the bottom type due to physical forces. 

The four bottom types have distinct physical and biological characteristics. Flat and rippled sand bottom types 
were composed primarily of sand substrate and secondarily shell rubble. Flat sand and rippled sand bottom 
types were characterized by low percent cover (0-2%) of benthic organisms at all sites. Although the sand bottom 
types were largely devoid of epifauna, numerous burrows indicate the presence of infaunal organisms. Sparse 
live bottom and ledges were colonized by macroalgae and numerous invertebrates, including coral, gorgonians, 
sponges, and "other" benthic species (such as tunicates, anemones, and bryozoans). Ledges and sparse live 
bottom were similar in terms of diversity (H') given the level of classification used here. However, percent cover 
of benthic species, with the exception of gorgonians, was significantly greater on ledge than on sparse live bot- 
tom. Percent biotic cover at sparse live bottom ranged from 0.7-26.3%, but was greater than 1 0% at only 7 out of 
51 sites. Colonization on sparse live bottom is likely inhibited by shifting sands, as most sites were covered in a 
layer of sediment up to several centimeters thick. On ledge bottom type, percent cover ranged from 0.42-1 00%, 
with the highest percent cover at ledges in the central and south-central region of GRNMS. 




Biotic cover on ledges is influenced by local ledge 
characteristics. Cluster analysis of ledge dimen- 
sions (total height, undercut height, undercut width) 
resulted in three main categories of ledges, which 
were classified as short, medium, and tall. Median 
total percent cover was 97.6%, 75.1%, and 17.7% 
on tall, medium, and short ledges, respectively. To- 
tal percent cover and cover of macroalgae, spong- 
es, and other organisms was significantly lower on 
short ledges compared to medium and tall ledges, 
but did not vary significantly between medium and 
tall ledges. Like sparse live bottom, short ledges 
may be susceptible to burial by sand, however the 
results indicate that ledge height may only be im- 
portant to a certain threshold. There are likely oth- 
er factors not considered here that also influence 
spatial distribution and community structure (e.g., 
small scale complexity, ocean currents, differential 
settlement patterns, and biological interactions). 




Image 1b. Large school of tomtates. 



GRNMS is a popular site for recreational fishing and boating, and there has been increased concern about the 
accumulation of debris in the sanctuary and potential effects on sanctuary resources. Understanding the types, 
abundance, and distribution of debris is essential to improving debris removal and education efforts. Approximate- 
ly two-thirds of all observed debris items found during the field surveys were fishing gear, and about half of the 
fishing related debris was monofilament fishing line. Other fishing related debris included leaders and spear gun 
parts, and non-gear debris included cans, bottles, and rope. The spatial distribution of debris was concentrated 
in the center of the sanctuary and was most frequently associated with ledges rather than at other bottom types. 
Several factors may contribute to this observation. Ledges are often targeted by fishermen due to the association 
of recreationally important fish species with this bottom type. In addition, ledges are structurally complex and are 
often densely colonized by biota, providing numerous places for debris to become stuck or entangled. Analysis of 
observed boat locations indicated that higher boat activity, which is an indication of fishing, occurs in the center 
of the sanctuary. On ledges, the presence and abundance of debris was significantly related to observed boat 
density and physiographic features including ledge height, ledge area, and percent cover. While it is likely that 
most fishing related debris originates from boats inside the sanctuary, preliminary investigation of ocean current 
data indicate that currents may influence the distribution and local retention of more mobile items. 

Fish communities at GRNMS are closely linked to benthic habitats. A list of species encountered, probability of 
occurrence, abundance, and biomass by habitat is provided. Species richness, diversity, composition, abun- 
dance, and biomass of fish all showed striking differences depending on bottom type with ledges showing the 
highest values of nearly all metrics. Species membership was distinctly separated by bottom type as well, al- 
though very short, sparsely colonized ledges often had a similar community composition to that of sparse live 
bottom. Analysis of fish communities at ledges alone indicated that species richness and total abundance of fish 
were positively related to total percent cover of sessile invertebrates and ledge height. Either ledge attribute was 
sufficient to result in high abundance or species richness offish. Fish diversity (FT) was negatively correlated with 
undercut height due to schools offish species that utilize ledge undercuts such as Pareques species. Concurrent 
analysis of ledge types and fish communities indicated that there are five distinct combinations of ledge type and 
species assemblage. These include, 1) short ledges with little or no undercut that lacked many of the undercut 
associated species except Urophycis earlii; 2) tall, heavily colonized, deeply undercut ledges typically with Ar- 
chosargus probatocephalus, Mycteroperca sp., and Pareques sp.; 3) tall, heavily colonized but less undercut 
with high occurrence of Lagodon rhomboides and Batistes capriscus; 4) short, heavily colonized ledges typically 
with Centropristis ocyurus, Halichoeres caudalis, and Stenotomus sp.; and 5) tall, heavily colonized, less under- 
cut typically with Archosargus probatocephalus, Caranx crysos and Seriola sp.. Higher levels of boating activity 
and presumably fishing pressure did not appear to influence species composition or abundance at the commu- 
nity level although individual species appeared affected. These results indicate that merely knowing the basic 



characteristics of a ledge such as total height, undercut width, and percent cover of sessile invertebrates would 
allow good prediction of not only species richness and abundance offish but also which particular fish species 
assemblages are likely to occur there. Comparisons with prior studies indicate some major changes in the fish 
community at GRNMS over the last two decades although the causes of the changes are unknown. 

Species of interest to recreational fishermen including Centropristis striata, Mycteroperca microlepis, and Myc- 
teroperca phenax were examined in relation to bottom features, areas of assumed high versus low fishing pres- 
sure, and spatial dispersion. Both Mycteroperca species were found more frequently when undercut height of 
ledges was taller. They often were found together in small mixed species groups at ledges in the north central 
and southwest central regions of the sanctuary. Both had lower mode size and proportion offish above the fish- 
ery size limit in heavily fished areas of the sanctuary (i.e. high boat density) despite the presence of better habitat 
in that region. Black sea bass, C. striata, occurred at 98% of the ledges surveyed and appeared to be evenly 
distributed throughout the sanctuary. Abundance was best explained by a positive relationship with percent cover 
of sessile biota but was also negatively related to presence of either Mycteroperca species. This may be due to 
predation by the Mycteroperca species or avoidance of sites where they are present by C. striata. 

Suggestions for monitoring bottom features, marine debris, and bottom fish at GRNMS are provided at the end of 
each chapter. The present assessment has established quantitative baseline characteristics of many of the key 
resources and use issues at GRNMS. The methods can be used as a model for future assessments to track the 
trajectory of GRNMS resources. Belt transects are ideally suited to providing efficient and quantitative assess- 
ment of bottom features, debris, and fish at GRNMS. The limited visibility, sensitivity of sessile biota, and linear 
nature of ledge habitats greatly diminish the utility of other sampling techniques. Ledges should receive the bulk 
of future characterization effort due to their importance to the sanctuary and high variability in physical structure, 
benthic composition, and fish assemblages. 




TABLE OF CONTENTS 

CHAPTER 1 JUSTIFICATION, OBJECTIVES AND APPROACH 1 

1.1 INTRODUCTION 1 

1.2 JUSTIFICATION 1 

1.3 METHODS 4 

Site Selection 4 

Field Methods 4 

REFERENCES 6 

CHAPTER 2 CHARACTERIZATION OF THE BENTHIC COMMUNITIES 7 

2.1 INTRODUCTION 7 

2.2 METHODS FOR BENTHIC SURVEYS 8 

2.3 METHODS FOR DATA ANALYSIS 10 

Thematic accuracy of benthic habitat maps 10 

Abiotic features 10 

Biotic cover 11 

Abiotic effects on biotic composition 12 

2.4 RESULTS 14 

Thematic accuracy 14 

Abiotic features 14 

Biotic cover 17 

Abiotic effects on biotic composition 20 

2.5 DISCUSSION 23 

2.6 RECOMMENDATIONS FOR MANAGEMENT AND MONITORING 33 

REFERENCES 35 

CHAPTER 3: CHARACTERIZATION OF THE MARINE DEBRIS 37 

3.1 INTRODUCTION 37 

3.2 METHODS FOR MARINE DEBRIS SURVEYS 38 

3.3 METHODS FOR DATA ANALYSIS 39 

Quantity, types, and spatial distribution of debris 39 

Effect of bottom type 39 

Influence of ledge characteristics and boat density on debris 39 

Predicting debris density 40 

Ocean currents at GRNMS 40 

3.4 RESULTS 41 

Quantity, types, and spatial distribution of debris 41 

Effect of bottom type 41 

Influence of ledge characteristics and boat density on debris 41 

Predicting debris density 44 

Ocean currents at GRNMS 44 

3.5 DISCUSSION 46 

3.6 RECOMMENDATIONS FOR MANAGEMENT AND MONITORING 50 

REFERENCES 52 

CHAPTER 4: CHARACTERIZATION OF THE BOTTOM FISH 55 

4.1 INTRODUCTION 55 

4.2 METHODS FOR FISH SURVEYS 56 

4.3 METHODS FOR DATA ANALYSIS 57 

4.4 RESULTS 59 

4.5 DISCUSSION 69 

4.6 TARGETED SPECIES 76 

4.7 RECOMMENDATIONS FOR MANAGEMENT AND MONITORING 78 

REFERENCES 80 

APPENDIX 83 

ACKNOWLEDGEMENTS 91 




List of Tables 

Table 1.1. Number of surveys within each bottom type 4 

Table 2.1. Variables measured to characterize benthic composition along fish transects 9 

Table 2.2. Error matrix for habitat classification from diver surveys at the GRNMS 12 

Table 2.3. A list of misclassified sites based on diver surveys 12 

Table 2.4. Spearman coefficients (rho) computed for pair-wise correlations among ledge dimensions 

for 92 ledge sites 14 

Table 2.5. Mean values and S.E. for dimensions of GRNMS ledge clusters determined from hierarchical 

clustering 14 

Table 2.6. Summary statistics for biotic composition by bottom types 16 

Table 3.1. Frequency of debris types pooled across all GRNMS survey sites 41 

Table 3.2. Presence and average number of debris items per 100m 2 by bottom type 43 

Table 3.3. Contrast estimates, standard errors, and chi-square statistics from logistic regression 

of presence of marine debris in GRNMS by bottom type 43 

Table 3.4. Two part conditional model for ledge bottom type 44 

List of Figures 

Figure 1.1. Location of Gray's Reef National Marine Sanctuary 2 

Figure 1.2. Spatial distribution of GRNMS bottom types and survey locations 3 

Figure 2.1. Schematic representation of the placement of the 1-m 2 -quadrat along a 25-m transect 

tape during fish and benthic substrate surveys at GRNMS 8 

Figure 2.2. Schematic representation of the physical ledge dimensions measured during benthic surveys. 9 

Figure 2.3. Stacked histogram plot of average abiotic substrate composition by substrate bottom type 13 

Figure 2.4. Ledge dimensions (total height, undercut height, undercut width) at ledge sites in GRNMS. ... 13 
Figure 2.5. Comparison of maximum ledge height measured in situ and determined through 

GIS analysis of sonar data 14 

Figure 2.6. Dendrogram produced from hierarchical clustering of ledge sites based on three mean 

ledge dimensions 14 

Figure 2.7. Three-dimensional plot of ledge clusters against ledge dimensions 15 

Figure 2.8. Histogram of average sand thickness on sparse live bottom sites 15 

Figure 2.9. Means of ledge variables in areas of high and low boat density 15 

Figure 2.10. Box plots of percent cover of benthic organisms on four bottom substrates at GRNMS 17 

Figure 2.11. Percent cover of biotic cover groups on flat and rippled sand bottom sites 18 

Figure 2.12. Percent cover of biotic cover groups at sparse live bottom sites 19 

Figure 2.13. Percent cover of biotic cover groups at ledge sites 20 

Figure 2.14. Box plots of number of gorgonians and sponges on four bottom substrates at GRNMS 21 

Figure 2.15. Box plots of height of gorgonians and sponges on four bottom substrates at GRNMS 21 

Figure 2.16. Box plots of Shannon diversity index (H) by bottom type 21 

Figure 2.17. Percent cover of coral types at ledge sites 22 

Figure 2.18. Percent cover of gorgonian types at ledge sites 23 

Figure 2.19. Percent cover of macroalgae types at ledge sites 24 

Figure 2.20. Percent cover of sponge types at ledge sites 25 

Figure 2.21. Percent cover of tunicates at ledge sites 26 

Figure 2.22. Percent cover of "other" benthic cover types at ledge sites 27 

Figure 2.23. Dendrogram from cluster analysis of all sites based on percent cover for aggregated 

cover types 28 

Figure 2.24. Mean percent cover of benthic organisms by clusters determined from hierarchal 

cluster analysis in Figure 2.23 29 

Figure 2.25. Dendrogram for cluster analysis of the 92 ledge sites based on percent cover for 

aggregated cover types 29 

Figure 2.26. Mean percent cover by cluster determined from hierarchal clustering of ledges 30 

Figure 2.27. Box plots of percent cover of benthic organisms on three ledge groups determined 

by cluster analysis 30 



Figure 


2.28. 


Figure 


2.29. 


Figure 


3.1a. 


Figure 


3.1b. 


Figure 


3.1c. 


Figure 


3.2. 


Figure 


3.3. 


Figure 


3.4. 


Figure 


3.5. 


Figure 


3.6. 


Figure 


3.7. 


Figure 


3.8. 


Figure 


3.9a. 


Figure 


3.9b. 


Figure 


3.10. 


Figure 


3.11. 


Figure 


3.12a 


Figure 


3.12b 


Figure 


3.13. 


Figure 


4.1. 


Figure 


4.2. 


Figure 


4.3. 


Figure 


4.4. 


Figure 


4.5. 


Figure 


4.6. 


Figure 


4.7. 


Figure 


4.8. 


Figure 


4.9. 


Figure 


4.10. 


Figure 


4.11. 


Figure 


4.12. 


Figure 


4.13. 


Figure 


4.14. 


Figure 


4.15. 


Figure 


4.16. 



Box plots of mean number of gorgonians and sponges on three ledge groups determined 

by cluster analysis of abiotic ledge variables 31 

Box plots of mean height of gorgonians and sponges on three ledge groups determined 

by cluster analysis 31 

Spatial distribution of total debris 42 

Spatial distribution of fishing gear 42 

Spatial distribution of non-gear 43 

Average number debris items per 100 m 2 transect by bottom type 44 

Frequency of debris at ledge sites by area and height class combinations 45 

Relationship of observed number of debris items with ledge height 45 

Relationship of observed number of debris items with percent cover of benthic organisms 45 

Relationship of observed number of debris items with undercut width 45 

Relationship of observed number of debris items with ledge area 45 

Locations of observed boats and density of boats per 0.25 km 2 cell 46 

Frequency histogram of the number of boats per cell in GRNMS 46 

Regions of low and high boat density 46 

For each 0.25 km 2 cell, the relationship of average number of debris items with the 

number of observed boats 47 

Interpolated debris density in GRNMS from ordinary kriging 48 

Frequency of current direction observations at 15m depth 49 

Tidal current direction overlaid on map of GRNMS 49 

Recommended strategy to prioritize ledge sites within GRNMS for debris removal 50 

Species richness, Shannon Diversity, abundance, and biomass offish within each 

bottom type 61 

Size frequency histograms offish by bottom type 62 

Dendrogram for cluster analysis of all 179 sites based on species composition 63 

Multiple regression model of species richness offish at ledge sites 64 

Multiple regression model of (log) abundance offish at ledge sites 64 

Multiple regression model of diversity of fish at ledge sites 65 

Dendrogram for cluster analysis of the 92 ledge sites based on species composition 66 

Dendrogram for cluster analysis of the 92 ledge sites based on ledge measurements 67 

Intersection of sites clustered based on ledge measurements and fish communities 68 

Size-frequency histograms for selected bottom fish targeted by the recreational fishery 69 

Size-frequency plots of Centropristis striata observed at ledge sites 70 

Size-frequency plots of Mycteroperca phenax observed at ledge sites 71 

Size-frequency plots of Mycteroperca microlepis observed at ledge sites 72 

Regression model of Centropristis striata abundance at ledge sites 73 

Logistic regression model of grouper species at ledge sites 74 

Logistic regression model of grouper species at ledge sites 75 




CHAPTER 1 JUSTIFICATION, OBJECTIVES AND APPROACH 

1.1 INTRODUCTION 

Baseline characterization of resources and patterns of use is an essential part of marine protected area (MPA) 
management (Kendall et al. 2004, Quattrini and Ross 2006). An inventory of the bottom features and the abun- 
dance, size structure, and habitat associations of organisms is needed to catalog and understand the environ- 
ment within the MPA. In addition, human use patterns must be understood to support development of manage- 
ment strategies designed to minimize user impacts. Once the MPAs resources and use patterns are adequately 
characterized, changes can be quantified and monitored through time. Such assessment and regular monitoring 
are critical to support adaptive management. Gray's Reef National Marine Sanctuary (GRNMS) currently lacks 
adequate characterization of several topics including bottom fish, benthic invertebrates, and human use patterns 
(NOAA2006). 

GRNMS is located on the inner continental shelf of the southeastern United States, 32.4 km offshore of Sapelo 
Island, Georgia (Figure 1.1, NOAA2006). The ecological importance of this area is related to its location at the 
transition between tropical and temperate waters, and the existence of a topographically complex system of 
ledges. The inner shelf area in the Georgia Bight is dominated by tidal currents and riverine runoff and is subject 
to seasonal variations in temperature, salinity and water clarity (Hanson et al. 1981; NOAA 2006). GRNMS is 
also influenced by the Gulf Stream along the outer shelf area, which transports deep nutrient-rich and temper- 
ate waters as well as tropical fish species to the area. Commonly referred to as "live bottom" areas, the rocky 
outcroppings within GRNMS support about 300 species of marine invertebrates (Gleason et al. 2005) and about 
65 species of macroalgae (Searles 1988). In turn, these benthic communities provide habitat for as many as 
1 50 fish species including several of interest to recreational and commercial fishermen (Sedberry and Van Dolah 
1984; Kennedy 1993). 

Prior to designation of the sanctuary in 1 981 , simple characterizations had been conducted in the area (e.g. Hunt 
1974). Since then, the sanctuary has been steadily building an inventory of its biological resources and human 
use patterns. In addition, field guides of the local fish (Gilligan 1989, Parker et al. 1994), algae (Searles 1988), 
and invertebrates (Gleason et al. 2005) have been produced. While limited in their spatial scope and quantitative 
rigor, these efforts have been instrumental in furthering knowledge of sanctuary resources. The recent comple- 
tion of detailed benthic maps of GRNMS (Figure 1 .2, Kendall et al. 2005) has led to a more robust and spatially 
explicit biological inventory and ecological characterization as reported here. Building on these earlier inventory 
activities, the present assessment was designed to address priority resource management issues as identified 
in the 2006 Final Management Plan for the sanctuary (NOAA 2006). This plan identified the management needs 
and goals for the next 5 years at GRNMS and specifically addressed the need for a quantitative, spatial charac- 
terization of marine debris, sessile invertebrates, benthic features, and the bottom fish associated with them. 




1.2 JUSTIFICATION 

Below are listed the GRNMS action plan items that 
were addressed by the present assessment fol- 
lowed by a description of how each item was met. 

STRATEGY MRP-3: REMOVE MARINE DEBRIS 
FROM THE SANCTUARY AND PREVENT NEW 
DEBRIS FROM ACCUMULATING 
Activity A: Clarify regulatory authority to address 
materials discharged or deposited outside the 
Sanctuary. 

Activity B: Develop and implement a marine debris 
education and outreach program. 
Activity C: Develop and implement a debris as- 
sessment and monitoring study. 
The present assessment addressed Activities B 
and C of this strategy. The assessment character- 



image 2. Diver studying benthic community with sponge species in the 
foreground. 




33°N- 



32°N- 



31°N- 



30°N- 



82°W 

L 



81 °W 

I 




-33°N 



-32°N 



Gray's Reef NMS 



n_n_r 

12.5 25 



~~ I Kilometers 

50 



-31 °N 



-30°N 



82°W 81 °W 

Figure 1.1. Location of Gray's Reef National Marine Sanctuary. 



80°W 



ized the type and quantity of marine debris at GRNMS, the type of bottom features it is routinely associated with, 
and proposed a strategy for continued assessment and monitoring. An understanding of the types and spatial 
distribution of marine debris in the sanctuary are necessary prerequisites to conduct clean-up and education ac- 
tivities efficiently. Knowledge of the spatial distribution of debris enables the sanctuary to focus clean-up efforts 
on the most affected areas given limited resources. Knowledge of the types of debris and potential mechanisms 
of transport and deposition allow education and outreach activities to focus their efforts on primary sources of 
debris. 



STRATEGY RM-3: ASSESS AND CHARACTERIZE SANCTUARY RESOURCES 
Activity A: Develop and update the GIS database. 
Activity B: Characterize benthic habitat. 
Activity C: Develop an invertebrate identification guide. 
Activity D: Develop the sanctuary characterization. 



The present assessment addresses Activities A, B, and D of the strategy. Bottom types were evaluated by sev- 
eral variables including biotic and abiotic cover types as well as ledge dimensions and sand characteristics. By 
linking the characterization of marine debris, benthic habitat, and bottom fish to spatial coordinates and the sea- 
floor map, the data can be easily incorporated into the GIS under development at the sanctuary. This will provide 
the most current, spatially explicit characterization of sanctuary resources available. 

STRATEGY RM-4: MAINTAIN AND ENHANCE MONITORING PROGRAMS 
Activity A: Monitor the status and health offish. 
Activity B: Design and implement an invertebrate monitoring program. 
Activity C: Develop a comprehensive water quality monitoring program. 
Activity D: Develop and implement a sediment analysis and monitoring program. 
Activity E: Support and enhance regional ocean observation systems. 
Activity F: Expand and update socioeconomic assessment. 
Activity G: Synthesize and characterize paleo-environmental information. 

The present assessment addresses Activities A and B of this strategy. Quantitative, spatially explicit charac- 
terization of fish communities and their associated habitats was a primary goal of this project. This includes 
species composition, size distribution, and density of each species by bottom type. In addition, recommenda- 




y — P?~*- s . 






??PS H_4£l-1] Kilometers 
'■~~€ 0.5"" -= t .1-o 



-V 



; :.£ 



m 



* A 



•%- r 



• v'-T^ •*'%•• 



\^. r 




■ r^m- • • • mr 




r . 



\:J 




Flat sand 
Rippled sand 



Sparsely colonized live bottom 
Densely colonized live bottom 
GRNMS Boundary 



• Survey locations 



Figure 1.2. Spatial distribution of GRNMS bottom types (classified by Kendall et al. 2005) and survey locations. 




tions for periodic assessment of bottom fish and 
sessile invertebrates are included to enable long 
term monitoring. 

The objectives of this characterization were to char- 
acterize the bottom fish, benthic features, marine 
debris, and the relationships among them for the 
different bottom types within the sanctuary: ledg- 
es, sparse live bottom, rippled sand, and flat sand. 
Particular attention was given to characterizing the 
different ledge types, fish communities of ledges, 
and marine debris associated with them given the 
importance of this bottom type to the sanctuary. 

1.3 METHODS 

Site Selection 

Field surveys were conducted in August 2004, May 
2005, and August 2005 to coincide with the avail- 
ability of a research vessel and favorable weather conditions. Since sampling periods were separated by only 
four months, seasonal differences in benthic communities were not explored in depth. Despite big differences in 
water temperature during May and August, even greater temporal separation of samples and more samples per 
year than used here would be better suited to characterizing seasonal differences. As a result, data are pooled 
for all analyses except where noted and the scope of inference is limited to the summer time period. 




Image 3. Jackknife fish, Equetus lanceolatus and coral. 





Aug 


May 


Aug 




Bottom Type 


2004 


2005 


2005 


Total Surveys 


Ledge 


15 


35 


42 


92 


Sparse live bottom 


17 


20 


14 


51 


Flat sand 


4 


8 


8 


20 


Rippled sand 


8 


5 


3 


16 



Most survey effort was devoted to the ledge bot- Table 1.1. Number of surveys within each bottom type. 
torn due to its high diversity and its importance to 
the sanctuary (Table 1.1). Less effort was devoted 
to the less diverse, lower complexity, and lower 
variability bottom types such as sparse live bottom 
and less still was devoted to the sand areas. Sites 
were selected randomly from within four bottom 
categories (flat sand, rippled sand, sparse live bot- 
tom, and densely colonized live bottom or ledges) identified in the recently completed benthic maps of GRNMS 
(Kendall et al. 2005). Sites in the sparse live bottom and sand categories were buffered such that they were a 
minimum of 30 m from other habitat types to ensure that a 25 m transect conducted along a random heading 
was contained within a single bottom type. Ledge sites are typically only a few meters wide, therefore surveys in 
that habitat were not conducted along a random heading. Ledge surveys were instead conducted along the lip 
of the ledge. Only ledges a minimum of 60 m long were allowed during site selection. A ledge 60 m long was the 
minimum size (+10 m) to accommodate a 25 m transect assuming that it was begun in the middle of the ledge 
and then conducted in a randomly chosen direction (i.e. left or right) along the ledge. If the random site selec- 
tion process resulted in a point on a ledge smaller than 60 m, the nearest ledge of suitable length was surveyed 
instead. However, this was not a common occurrence, as <5% of the randomly selected sites were located on 
ledges of unsuitable length. Number of surveys by bottom type and sampling period are provided in Table 1.1, 
and the spatial distribution of survey locations is shown in Figure 1 .2. 

Field Methods 

There were three components to the field survey: fish counting, benthic assessment, and quantification of marine 
debris, all of which occurred within a 25 x 4 m belt transect for a total survey area of 1 00 m 2 . Two divers surveyed 
the transect at each survey site. One diver was responsible for visual counts and size estimation of fish species. 
The second diver characterized benthic features with five randomly placed 1 m 2 quadrats and quantified marine 
debris within the entire transect. More details on field methods will be provided in subsequent sections. 

For all bottom types except ledge, the divers selected a random compass heading (0-360°) along which to con- 
duct the survey. Exceptions were made at sites with a strong current or surge, where the survey was conducted 



into the current to ease the physical demands on 
divers. Surveys over ledge habitat were conducted 
along the ledge face or lip (if undercut) and fol- 
lowed any turns or curves along the ledge. This 
ensured that the entire survey would be conducted 
within the ledge bottom type. Once at a site, the 
fish surveyor attached a tape measure to the sub- 
strate or weighted line that was used to mark the 
site and began the survey. The entire length of the 
transect survey was conducted at a relatively con- 
stant speed and fixed time period (~1 5 minutes) re- 
gardless of bottom type or number of fish present. 
Detailed survey methods and protocols for data 
collection for bottom fish, benthic assessment, and 
quantification of marine debris are provided in the 
subsequent chapters respectively. 




Image 4. Diver conducting habitat survey. 




REFERENCES 

Gilligan, MR. 1989. An Illustrated Field Guide to the Fishes of Gray's Reef National Marine Sanctuary. NOAA Technical 
Memorandum, NOS MEMD 25. Marine and Estuarine Management Division, OOCRM, NOS, NOAA, U.S. Department of 
Commerce, Washington, D.C. February 1989. 77 p. 

Gleason, DF, AW Harvey, SP Vives. 2005. A guide to the benthic invertebrates and cryptic fishes of Gray's Reef. Georgia 
Southern University. Statesboro GA. http://www.bio.georgiasouthern.edu/GR-inverts/ 

Hanson, RB, KR Tenore, S Bishop, C Chamberlain, MM Pamatmat, and J Tietjen. 1 981 . Benthic enrichment in the Georgia 
Bight related to Gulf Stream intrusions and estuarine outwelling. Journal of Marine Research 39(3): 417-441. 

Hunt, JL. 1 974. The geology and origin of Gray's Reef, Georgia continental shelf. Master's Thesis, Athens, Georgia. 

Kendall, MS, JD Christensen, C Caldow, M Coyne, C Jeffrey, ME Monaco, W Morrison Z Hillis-Starr. 2004. The influence of 
bottom type and shelf position on biodiversity of tropical fish inside a recently enlarged marine reserve. Aquatic Conserva- 
tion: Marine and Freshwater Ecosystems 14:113-132. 

Kendall, MS, OP Jensen, C Alexander, D Field, G McFall, R Bohne, and ME Monaco. 2005. Benthic mapping using sonar, 
video transects, and an innovative approach to accuracy assessment: A characterization of bottom features in the Georgia 
Bight. Journal of Coastal Research 21(6): 1154-1165. 

Kennedy, J. 1993. Gray's Reef National Marine Sanctuary: after a decade in the sanctuary program, it still has secrets to 
share. Mariner's Weather Log 37:2a-8a. 

NOAA. 2006. Gray's Reef National Marine Sanctuary Final Management Plan/Final Environmental Impact Statement. 
NOAA NOS NMSP, Savannah, GA. 



Parker Jr., RO, AJ Chester, and RS Nelson. 1994. Avideo transect method for estimating reef fish abundance, composition, 
and habitat utilization at Gray's Reef National Marine Sanctuary, Georgia. Fishery Bulletin 92:787-799. 

Quattrini, AM, SW Ross. 2006. Fishes associated with North Carolina shelf-edge hardbottoms and initial assessment of a 
marine protected area. Bulletin of Marine Science 79:137-163. 

Searles, RB. 1988. An Illustrated Field Guide to the Seaweeds of Gray's Reef National Marine Sanctuary. NOAA technical 
memorandum NOS MEMD 22, Silver Spring, MD. 

Sedberry, GR, and RF Van Dolah. 1984. Demersal fish assemblages associated with hard bottom habitat in the South At- 
lantic Bight of the USA. Environmental Biology of Fishes 11(4): 241-258. 



CHAPTER 2 CHARACTERIZATION OF BENTHIC COMMUNITIES 

2.1 INTRODUCTION 

Morphologically complex hardbottom outcroppings are interspersed throughout the continental shelf from the 
coast to the shelf edge along the South Atlantic Bight (Parker et al. 1983; Van Dolah et al. 1994). These rocky 
features vary from flat, smooth surfaces to exposed vertical scarps and ledges with numerous overhangs, crev- 
ices, and slopes (Riggs et al. 1996). Exposed surfaces are colonized to varying extents by algae and sessile 
and burrowing invertebrates, which in turn provide shelter, foraging, and nursery areas for a large diversity of 
fish. In addition to providing important habitat, hardbottom substrate is an important source of sand production 
on sediment-starved areas on the continental shelf (Riggs et al. 1998). Although several studies have examined 
the distribution (Henry and Giles 1979; Parker et al. 1983; Van Dolah et al. 1994) and geological origins (Riggs 
et al. 1996) of hardbottom habitats in the South Atlantic Bight, less is known about the benthic flora and fauna 
communities that inhabit these substrates (but see Wenner et al. 1983; Peckol and Searles 1984). 

Gray's Reef National Marine Sanctuary (GRNMS) encompasses approximately 58 km 2 , about 75% of which is 
comprised of unconsolidated sediments, including flat sand plains and rippled sand (Kendall et al. 2005). The 
remaining substrate consists of outcroppings of carbonate hardbottom formed during the Pliocene era. The 
hardbottom ranges from areas with little or no vertical relief to areas of irregular, high-relief rocky ledges (> 2 m) 
where invertebrate growth is abundant (Henry and Giles 1979; Van Dolah et al. 1994; Kendall et al. 2005). The 
vast majority (-97%) of the hardbottom areas at GRNMS are flat, contain a thin veneer of sand overlying sand- 
stone or limestone rock, and are sparsely colonized by sessile invertebrates. Densely colonized ledges account 
for <1 % of the total bottom. 



A recent review of the GRNMS management plan identified the need to assess and characterize sanctuary re- 
sources to understand the associations among biological, physical and geological components of the ecosystem 
being protected (NOAA2006). To this end, one focus of sanctuary management has been to classify and charac- 
terize benthic habitats through mapping activities (Kendall et al. 2005). Benthic maps provide an understanding 
of the spatial distribution of benthic habitats within GRNMS and a spatial framework for addressing research and 
management questions such as identifying and protecting essential fish habitats. The premise here is that strong 
linkages exist between fishes and their habitats and these linkages affect the spatial distribution of these ani- 
mals. However, benthic maps may be somewhat limited because they comprise a mosaic of habitat patches that 
are static point-in-time estimates of dynamic properties of the ecosystem they represent (Weins 1976). In addi- 
tion, habitat patches are assumed to be areas of homogeneous ecological and environmental conditions with 
discrete boundaries at a specific spatial and temporal scale. For example, all map features labeled as densely 
colonized ledges are assumed to be identical at the mapped scale but may contain complex spatial patterns in 
the distribution of resources that are not only variable when examined at the finer scales but also are temporally 
dynamic. Finer-scale in situ assessments and characterizations of benthic substrates are needed to quantify 1) 
the accuracy of hardbottom delineations, and 2) within and between substrate variability that may be affecting 

benthic communities and fish assemblages within 
GRNMS (Kendall et al. 2005). 

In addition, the recent management plan calls for 
activities to maintain and enhance monitoring pro- 
grams (NOAA 2006). At GRNMS, invertebrates 
comprise the most diverse, abundant, and conspic- 
uous component on hardbottom habitats, but pre- 
vious assessments and monitoring attempts have 
not yielded appropriate data to detect changes 
in abundance, density or presence/absence over 
time (NOAA 2006). To fill this data gap, the current 
study provides a consistent and comprehensive 
in situ assessment of benthic communities, which 
can be used to design, implement, and maintain 
an invertebrate monitoring program at GRNMS. 
Moreover, this study provides data that will help 
quantify detailed habitat-fish associations, which 
will be useful in developing quantitative estimates 




Image 5. A close-up look at a diverse benthic community. 




of habitat utilization by fishes and provide the spatial framework 
needed to address management goals relative to protecting es- 
sential fish habitats. 

Finally, data collected during this study will be used to evaluate 
independently the thematic accuracy of the GRNMS benthic 
map produced by Kendall et al. (2005). Kendall et al. (2005) 
evaluated the accuracy of the benthic habitat map with video 
transect data, which showed a very high degree of thematic 
accuracy. However, due to limitations of the accuracy assess- 
ment data set, they were restricted to evaluating only two of 
the four bottom types, sparsely colonized live bottom and un- 
consolidated sediment. Utilizing in situ classifications of habitat 
type will enable us to evaluate the accuracy of all four bottom 
types, including ledges. 

This study provides baseline estimates of the composition of 
four mapped benthic substrates within GRNMS as identified by 
Kendall et al. (2005). The overall goal of this study was to pro- 
vide detailed and complete assessments of benthic substrates 
within GRNMS, especially the ledge bottom types. Specific ob- 
jectives were as follows: 




Image 6. Diver measuring ledge dimensions. 



1. Evaluate independently the thematic accuracy of the 
GRNMS benthic map produced by Kendall et al. (2005); 

2. Characterize the abiotic features of the benthos within each mapped bottom substrate; 

3. Characterize the types, distribution, abundance of benthic flora and fauna within mapped substrates; and, 

4. Identify abiotic features that may be influencing the spatial distribution, composition and abundance of inver- 
tebrate assemblages at ledges. 

2.2 METHODS FOR BENTHIC SURVEYS 

Using the same classification scheme that was used in benthic maps (Kendall et al., 2005), the diver indepen- 
dently assigned an overall bottom type to each transect based on in situ observation. Bottom types were ledge, 
sparse live bottom, flat sand, or rippled sand. The bottom type was assigned independently of that expected 
based on the benthic map so that the in situ data could be used for map validation. Data on the percent cover 
of abiotic and biotic composition at each survey site were recorded within five 1 m 2 quadrats along the 25 x 4 m 
transect. Some sites in August 2005 had only four quadrats evaluated due to scuba diving time limits. The quad- 
rat was placed at each randomly chosen meter mark and systematically alternated from side to side along the 
transect tape, except on ledges (Figure 2.1 ). When characterizing the narrow ledges, the quadrats were placed 
entirely on the ledge rather than on the often bare substrate below it. It is important to note that beyond the scarp 
and first 1-2 meters of the top of the ledge, the bottom transitions into sparse live bottom. Transects at ledge 
sites were conducted solely along this ledge edge and not the sparse live bottom behind it. At all sites, several 
variables were measured to characterize benthic composition and structure (Table 2.1 ). The quadrat was divided 
into 100 smaller 10 x 10 cm squares with string (1 small square = 1% cover) to help the diver with estimation of 



4 m 



□ 



□ 



?□ 



10 



□ 15 



Transect tape 



20 □ 

1-m 2 quadrat 



25 m 



Figure 2.1. Schematic representation of the placement of the 1-m 2 -quadrat along a 25-m transect tape during fish and benthic sub- 
strate surveys at GRNMS. Broken line represents the total area surveyed (100 m 2 ). 




Table 2.1. Variables measured to characterize benth 


c composition along fish transects. 




Benthic Composition 


% Cover 


He 


ight (cm) 


Abundance (#) 


Abiotic 


Hard 


X 








Sand 


X 








Shell rubble 


X 








Fine sediment 


X 








Biotic 


Corals 










Branching 


X 








Cup 


X 








Ball 


X 








Encrusting 


X 








Other 


X 








Gorgonians 










Sea rods/ plumes 


X 




X 


X 


Sea fans 


X 




X 


X 


Sea whips 


X 




X 


X 


Macroalgae 










Filamentous /Turf 


X 








Fleshy 


X 








Other 


X 








Sponges 










Ball 


X 




X 


X 


Vase 


X 




X 


X 


Tube 


X 




X 


X 


Finger 


X 




X 


X 


Rope 


X 




X 


X 


Encrusting 


X 








Other benthic macrofauna 










Anemones 


X 








Tunicates (Encrusting) 


X 








Tunicates (Lobate) 


X 








Zoanthids 


X 








Other 


X 









percent cover. Percent cover was determined by looking at the quadrat from above and visually estimating per- 
cent cover in a two dimensional plane. The percent cover (to the nearest 1 %) of four abiotic substrate categories 
was determined first. The categories of abiotic substrates were hardbottom, sand, shell rubble, and fine sedi- 
ments (Table 2.1). Hardbottom referred to consolidated substrates including those that were covered in a thin 
veneer of sand less than 10 cm thick and immovable, consolidated shell rock substrates. Shell rubble referred 
to loose shells or shell fragments that were moveable. Fine sediments were substrates consisting of unconsoli- 
dated silt that was easily resuspended and remained in the water column. 



Percent cover (to the nearest 0.1 %) of the sessile biota was also determined for major taxonomic groups, which 
were further subdivided into categories based on morphology (Table 2.1). The maximum height and number of 
individual colonies were recorded for each sponge 
and gorgonian morphology. Uncolonized substrata 
were recorded as bare substrate. 



Other specific measurements were made on ledge, 
sparse live bottom, and sandy bottom types. For 
ledge sites, the dimensions of ledges were record- 
ed at each quadrat position. Total height was mea- 
sured from the base of the ledge to the top of the 
substrate behind it but excluded the height of ses- 
sile organisms that were attached to the substrate 
(Figure 2.2). Undercut width - the distance from 



Undercut height 



Undercut width 




Total ledge 
height 



Figure 2.2. Schematic representation of the physical ledge dimensions 
measured during benthic surveys. 




HI 



the leading face of the ledge to the farthest recess 
under the ledge - was visually estimated either 
by using the tape as a reference or by inserting 
the quadrat under the ledge (Figure 2.2). Under- 
cut height - the height under the ledge - also was 
estimated visually with the length of the quadrat 
as a reference (Figure 2.2). Undercut width and 
height were recorded as zero for ledges that were 
not undercut. On sparse live bottom, the depths of 
sand, shell rubble, or fine sediment (hereafter sand 
thickness) were measured from the surface to the 
underlying limestone bottom up to 30 cm deep. On 
flat and rippled sand, the number of holes from bur- 
rowing organisms that occurred within each quad- 
rat was recorded. In addition, on rippled sand, the 
wavelength and height of sand ripples within each 
quadrat were recorded. 

Image 7. Example of sand bottom in GRNMS. 
Sites were used as independent sample units and 

were considered replicates within each bottom type. Multiple quadrat measurements within each site transect 
were averaged using the equation: HQ^J/n, where Q. = quadrat i, and n = total number of quadrats. Average site 
values were then used to calculate means and standard errors of measured variables for each bottom type. 

2.3 METHODS FOR DATA ANALYSIS 
Thematic accuracy of benthic habitat maps 

Thematic accuracy of the benthic map of GRNMS was estimated using diver surveys at all 179 locations. Re- 
sults including overall accuracy, user's and producer's accuracy, and Kappa statistic (Congalton and Green 
1999) are summarized in an error matrix for all four mapped bottom types: flat sand, ledge, rippled sand, and 
sparse live bottom. 




Abiotic features 

The mean percent cover of each abiotic substrate category was plotted by bottom type. The hypotheses that 
1) the amount of hardbottom varied significantly between ledges and sparse live bottom and 2) the amount of 
sand and shell rubble varied significantly among flat sand and rippled sand were tested non-parametrically with 
a Wilcoxon Rank Test (JMP v5.1 ). 

The physical dimensions (total height, undercut height, and undercut width) at all ledge sites were entered into 
a geographic information system (GIS) and mapped to examine any spatial patterns of these ledge character- 
istics in the sanctuary. In addition, Spearman's Rho was computed on the ranks of data values to determine if 
significant pair-wise correlations existed among ledge dimension variables. Rho is a non-parametric correlation 
parameter that ranges from 1 to -1 with a value of 1 indicating a strong positive relationship and -1 indicating 
a strong negative relationship between paired variables (Sokal and Rohlf 1995). Then hierarchical clustering 
(Ward's minimum variance method) was used to identify groups of ledge sites with similar in situ physical dimen- 
sions (JMP v5.1). The stability of the clusters was checked by running the cluster analysis with multiple fusion 
strategies. The sites and their corresponding clusters were plotted against total height, undercut height, and 
undercut width. 

Estimates of ledge height for all ledges within GRNMS were derived previously from a GIS analysis of sonar 
bathymetry by Kendall and Eschelbach (2006). Briefly, they determined ledge height for each polygon using 
a 2 meter resolution bathymetry grid of the sanctuary. All depth values from the bathymetry grid around each 
polygon were extracted from the bathymetry data and the deepest and shallowest values were subtracted to 
determine the maximum depth change or height for each ledge. To assess the accuracy of the GIS height esti- 
mates, the maximum ledge height determined by Kendall and Eschelbach (2006) was compared with maximum 
ledge height recorded in situ. In situ estimates of ledge height from the present study were plotted against GIS 




estimates from Kendall and Eschelbach (2006) and a nonparametric Spearman's Rho correlation was computed 
on the ranks of the data. 




.*. 




Image 8. Sparsely colonized live bottom (with bat fish) in GRNMS. 



A frequency histogram was used to determine the 
distribution of data on sand thickness from sparse 
live bottom sites. Data on sand characteristics 
(sand ripple height and wavelength) were calcu- 
lated for rippled sand. The number of burrows on 
flat and rippled sand was compared with a one- 
way ANOVA test. 

Finally, we examined how ledge characteristics 
differ across regions of varying human use. Dis- 
cussions with recreational fishermen indicated 
that they tend to target ledges that are tall and 
large in area. Indeed, larger and taller ledges also 
contained significantly greater amounts of ma- 
rine debris, which is additional evidence of fishing 
(Chapter 3). Several ledge metrics, such as area 
(measured through GIS analysis) ledge height, un- 
dercut height, undercut width, and percent cover 
of benthic organisms, were therefore quantified 
and compared between areas of high and low boat 
density, as identified in Chapter 3, using paramet- 
ric one-way ANOVA tests (JMP v5.1). 

Biotic cover 

The percent cover of each biotic category was plotted by bottom type and interquantile (25 th , 75 th ) statistics were 
calculated. The relative total percent cover was the sum of all quadrat measurements (converted to a percent- 
age) of each biotic category across all sites within each bottom type. The hypotheses that the biotic cover varied 
among bottom types was tested non-parametrically with Kruskal-Wallis Tests (JMP v5.1) and Dunn's multiple 
comparison tests. Similarly, height and abundance of sponges and gorgonians were plotted by bottom type and 
interquantile (25 th , 75 th ) statistics were calculated. The hypothesis that height and number of individuals varied 
among bottom type was tested non-parametrically with Kruskal-Wallis Tests (JMP v5.1) and Dunn's multiple 
comparison tests. 

Percent cover of main cover types at all sites was 
mapped in a geographic information system (GIS). 
Pie chart symbology was used to depict percent 
cover of corals, gorgonians, macroalgae, sponge, 
"other", and bare substrate at each site. Loca- 
tion of sites was occasionally adjusted slightly to 
prevent chart overlap. In addition, detailed maps 
were made for ledge bottom to show percent cov- 
er of subcategories within each main cover type. 
For each main cover type, bar charts were used 
to depict percent cover of subcategories at each 
ledge site. Two maps of this type were made for 
the "other" category. Due to the importance of tuni- 
cates as a sanctuary resource, the subcategories 
devoted to tunicates (encrusting and lobate) were 
mapped separately to better discern spatial pat- 
terns of tunicate cover. The second map included 
the remaining subcategories (anemones, benthic 
dwelling zoanthids, and other). 




Image 9. Short ledge with 1x1 m quadrat. 




HI 



Shannon-Weiner diversity (H') of biotic cover for 
each site was calculated using the detailed (non- 
aggregated) morphology types as: 

x 

H'=-I Pi ln( Pj ) 



where x is the total number of morphological types 
(see Table 2.1 for list of all types) and p. is the pro- 
portion of area covered by morphological type i. 
The hypothesis that diversity varied among bottom 
types was tested non-parametrically with a Krus- 
kal-Wallis Test (JMP v5.1) and Dunn's multiple 
comparisons tests. 

Biotic data were also analyzed by hierarchal clus- 
tering using Ward's minimum variance method. 
The percent cover variables, aggregated by cover 
type (corals, gorgonians, macroalgae, sponge, 
and other), were the basis for the analysis. A scree 
plot was used to help determine 
the number of well-separated 
clusters from the dendrogram. 
The stability of the clusters was 
checked by running the cluster 
analysis with multiple fusion 
strategies. In addition, the clus- 
tering procedure was conducted 
for ledges only using the same 
methods. 




Image 10. Example of a tall, undercut ledge in GRNMS. A 1x1 m quad- 
rat is shown for scale. 



Table 2.2. Error matrix for habitat classification from diver surveys at the Grays' Reef National 
Marine Sanctuary. U = user's accuracy and P = producer's accuracy. 

Diver assessed 
bottom type 



Abiotic effects on biotic 
composition 

Finally, the relationship be- 
tween the benthic community 
structure and abiotic ledge 
characteristics was assessed. 
Graphical methods and non- 
parametric tests were utilized 
to examine whether the percent 
cover, number of individuals, 
and height of benthic organ- 
isms varied among the three 
ledge categories (short, medi- 
um, and tall), which were deter- 
mined from the cluster analysis 
of abiotic variables (see Re- 
sults- Abiotic features). Percent 
cover of each biotic type (total, 
coral, gorgonians, macroalgae, 
sponge, and other organisms) 
was plotted by ledge category 
and interquantile (25 th , 75 th ) 
statistics were calculated. The 



Mapped 
bottom type 



Count 


Flat sand 


Ledge 


Rippled sand 


Sparse live 
bottom 


Total 


Flat sand 


18 
100% (U) 
85.7% (P) 











18 


Ledge 





92 

91.1% (U) 
100 %(P) 


2 


7 


101 


Rippled sand 


3 





13 
81.3% (U) 
86.7% (P) 





16 


Sparse live 
bottom 











44 

100 %(U) 
86.3 % (P) 


44 


Total 


21 


92 


15 


51 


179 



"Overall accuracy = (167/179)*100% = 93.3% 
Kappa = 0.89 ± 0.03 



Table 2.3. A list of misclassified sites based on diver 


surveys (n = 12). 






Mapped 




Diver Assessed 




Site 


Date 


Bottom 


Type 


Bottom Type 


Diver Notes 


D19 


Aug 04 


Ledge 




Rippled sand 


change in depth, sparse, 
surveyed sand nearby 


D22 


Aug 04 


Ledge 




Sparse live bottom 


None 


D23 


Aug 04 


Ledge 




Sparse live bottom 


small ledge nearby 


D27 


Aug 04 


Ledge 




Rippled sand 


change in depth, surveyed 
rippled sand nearby 


D30 


Aug 04 


Ledge 




Sparse live bottom 


None 


D4 


Aug 04 


Ledge 




Sparse live bottom 


None 


D42 


Aug 04 


Ledge 




Sparse live bottom 


None 


D8 


Aug 04 


Ledge 




Sparse live bottom 


small ledge nearby 


D9 


Aug 04 


Ledge 




Sparse live bottom 


None 


RS1 


Aug 05 


Rippled 


sand 


Flat sand 


None 


RS7 


Aug 05 


Rippled 


sand 


Flat sand 


many echinoderms, >15 


RS9 


Aug 05 


Rippled 


sand 


Flat sand 


None 



hypothesis that the percent cover 
varied among ledge categories 
was tested non-parametrically with 
Kruskal-Wallis Tests (JMP v5.1 ) and 
Dunn's multiple comparison tests. 

Similarly, height and abundance of 
sponges and gorgonians were plot- 
ted by ledge category and interquan- 
tile (25 th , 75 th ) statistics were calcu- 
lated. The hypothesis that height and 
number of individuals varied among 
ledge category was tested non- 
parametrically with Kruskal-Wallis 
Tests (JMP v5.1) and Dunn's multi- 
ple comparison tests. The mean and 
standard error of Shannon diver- 
sity was calculated for each ledge 



> 
O 

u 

<D 
O 

i_ 

<D 
Q_ 

C 
TO 
<D 



100% 

90% - 
80% - 
70% - 
60% - 
50% 
40% 
30% 
20% 
10% -I 
0% 



n=92 



n=51 



n=21 



n=15 



■ Shell rubble 

□ Sand 

□ Hard bottom 



Ledge Sparse live Flat sand Rippled 

bottom sand 

Figure 2.3. Stacked histogram plot of average abiotic substrate composition (relative 
total percent cover) by substrate bottom type. 



*■• ?m*- s . 




~i jf^"1 Kilometers 
~-iT 0.5 * ' - v 1 '-'I- 



'A' 



Flat sand 
Rippled sand 



Sparsely colonized live bottom 
Densely colonized live bottom 
GRNMS Boundary 



Ledge dimensions (cm) 



J 



150 



Mean total height 
Mean undercut height 
Mean undercut width 



Figure 2.4. Ledge dimensions (total height, undercut height, undercut width) at ledge sites in GRNMS. The tallest bar in the legend 
represents 150 cm. 




300 -, 



250 - 



S. 200 
"5 150 



« 100 



Rho = 0.63 
p <0.0001 



/ 



/ ♦ 



♦ 
♦ ♦ 



50- #>*♦ «♦ 







50 






100 150 200 250 300 

GIS height (cm) 

Figure 2.5. Comparison of maximum ledge 
height measured in situ and determined through 
GIS analysis of sonar data. The red dashed line 
represents the theoretical 1:1 ratio. Statistics for 
Spearman's rank correlation are provided. 

height category and the hypothesis that di- 
versity varied between ledge height class- 
es was tested using a one-way ANOVA test 
(JMPv5.1). 

2.4 RESULTS 
Thematic accuracy 

Overall map accuracy was quite high 
(93.3%), with 167 of 179 sites being cor- 
rectly classified (Table 2.2). Kappa was 
0.89 ± 0.03 indicating that the classification 
in the map is -89% better than that expect- 
ed if bottom types were randomly assigned 
to each polygon. A user of the benthic map 
can expect nearly 100% accuracy for flat 
sand and sparse live bottom, 81.3% for 
rippled sand, and 91.1% for ledge. Three 
sites classified in the map as rippled sand 
were identified as flat sand by divers. Nine 
of the sites classified as ledges in the map 
were identified as other bottom types by 
divers. Two such ledge sites were identified 
as rippled sand and the other seven were 
identified as sparse live bottom. A list of all 
misclassified sites along with relevant diver 
notes is given in Table 2.3. 

Abiotic features 

Ledge and sparse live bottom sites were 
dominated by hardbottom substrates, with 
very small amounts of sand or shell rubble 
cover (Figure 2.3). Mean cover of hardbot- 
tom did not vary significantly between ledge 
and sparse live bottom sites (X 2 = 0.5991, 
df = 1, p = 0.44). No hardbottom was ob- 
served at either flat or rippled sand sites 



Table 2.4. Spearman coefficients (rho) computed for pair-wise correlations 
among ledge dimensions for 92 ledge sites. 



Variable 



by Variable 



Rho 



Pr>Rho 



Undercut width (cm) 
Ave. Undercut height (cm) 
Ave. Undercut height (cm) 



Total height (cm) 0.7029 

Total height (cm) 0.7269 

Undercut width (cm) 0.9673 



<0001 
<0001 
<0001 



Table 2.5. Mean values and S.E. for dimensions of GRNMS ledge clusters 
determined from hierarchical clustering (Ward's minimum variance). 
Cluster N Group Total Height Undercut Width Undercut Height 

1 60 Short 12.3(1.0) 3.0(0.7) 1 .4 (0.3 

2 26 Medium 45.5(2.9) 34.0(6.9) 15.5(1.3) 

3 6 Tall 115.8(18.9) 175.1(38.5) 38.4(2.1) 



b 



Cluster 1 

(Short) 



i^- 



0- 



>> 



Cluster 2 

(Medium) J 



Cluster 3 

(Tall) 




Q Lwv>«««~w 



Figure 2.6. Dendrogram produced from hierarchical clustering of ledge 
sites based on three mean ledge dimensions (total height, undercut 
height, and undercut width). 




(Figure 2.3). Sand and shell rubble dominated flat 
and rippled sand bottom sites (Figure 2.3). Shell 
rubble generally occurred in the troughs of sand 
waves on rippled sand sites. Ranking of mean 
sand and shell rubble values was not significantly 
different between flat sand and rippled sand sites 
(X 2 = 0.56, df = 1 , p = 0.45). Fine sediment was not 
observed at any site surveyed. 

The physical dimensions of the ledges surveyed 
exhibited wide variation and did not exhibit distinct 
spatial patterns (Figure 2.4). Strong positive pair- 
wise correlations occurred among ledge dimen- 
sions such that an increase in total height also 
correlated with an increase in undercut height and 
undercut width (Table 2.4). In particular, undercut 
height was highly correlated with undercut width 
(Table 2.4) Total height explained 70 and 72% of 
the variability in average undercut width and height, 
respectively. In situ maximum height correlated 
with maximum ledge height determined from the 
GIS analysis of sonar bathymetry (rho = 0.63, p < 
0.0001). However, with the exception of one case, 
GIS derived heights were always higher than in situ 
field measurements (Figure 2.5). 

Hierarchical clustering identified three groups of 
ledge sites based on their physical dimensions (Fig- 
ure 2.6). Cluster one (hereafter "short") was the larg- 
est group and contained 60 short ledge sites with 
little or no undercut (Figure 2.6, Table 2.5). Cluster 
two (hereafter "medium") contained 23 ledge sites 
that were of medium height and moderate undercut, 
while Cluster 3 (hereafter "tall") had six tall ledges. 
A three dimensional plot of ledge sites against their 



Mean total height (cm) 
170 






44.0 

29.3 

Mean undercut height (cm) 



197 
Mean undercut width (cm) 



Figure 2.7. Three-dimensional plot of ledge clusters against ledge di- 
mensions. Ledge sites from GRNMS were classified into three groups 
by hierarchical clustering using the Ward's minimum variance method. 



16- 



14- 



12- 



10- 




0- 



Frequency 

Figure 2.8. Histogram of average sand thickness on sparse live bottom 
sites. 



3000 

2500 

, 2000 

' 1500 

1000 

500 



14 

I 12 
£ 10 

~ 8 

5 6 

"O 

§ 4 
c 

<y 2 
E 




F Ratio = 11.97, 
p = 0.0008 



High 



- 






F Ratio = 5.0, 
p = 0.0279 
































- 









High 



50 

_ 45 - 
| 40 
~ 35 

■130 

#20- 

c 15 

S 10 

5 5 



45 
? 40 
^35 

£1 

S 30 
S 

- 25 

S 20 

T3 

«= 15 

=! 

S 10 



High 




High 



F Ratio = 9.01, 
p = 0.0035 



fl_ 



F Ratio = 2.22, 
p = 0.1395 



■_ 



70 

Co 60 

5 50 

> 

o 

« 40 

c 

a> 

S£ 30 

<u 

a. 

c 20 

<u 

E 10 



F Ratio = 16.0, 
p = 0.0001 



High 



Figure 2.9. Means (± SEM) of ledge variables in areas of high and low boat density. Results of one-way ANOVA tests are provided 
(df = 91,a = 0.05). 




Table 2.6. Summary statistics for biotic composition by bottom types. Blank cells indicate that zero organ- 
isms were observed. 



COVER TYPE 


Morphology 


Biotic variable 


Ledge 


Sparse 


live bottom 


Flat 


sand 


Rippled 


sand 


Mean 


S.E. 


Mean 


S.E. 


Mean 


S.E. 


Mean 


S.E. 


Corals 


Ball 


% Cover 










Branching 


% Cover 


1.2 


0.2 


<0.1 


<0.1 






Cup 


% Cover 


<0.1 


<0.1 


<0.1 


<0.1 






Encrusting 


% Cover 


<0.1 


<0.1 


<0.1 


<0.1 






Other 


% Cover 


0.1 


0.1 


<0.1 


<0.1 






Gorgonians 


Sea fans 


# Individuals 
% Cover 
Ht (cm) 


<0.1 
<0.1 
1.4 


<0.1 
<0.1 
0.5 


<0.1 
<0.1 
1.1 


<0.1 
<0.1 
0.4 






Sea rod/plume 


# Individuals 
% Cover 
Ht (cm) 


3.3 
1.3 
13.5 


0.4 
0.2 
1.4 


4.9 
1.5 
19.4 


0.5 
0.2 
1.5 


<0.1 
<0.1 
0.7 


<0.1 
<0.1 
0.8 


<0.1 
<0.1 
1.2 


<0.1 
<0.1 
0.9 


Sea whips 


# Individuals 
% Cover 
Ht (cm) 


0.1 
<0.1 
2.0 


<0.1 
<0.1 
0.6 


0.1 
<0.1 
3.6 


<0.1 
<0.1 
1.1 


<0.1 
<0.1 
<0.1 


<0.1 
<0.1 
<0.1 




Macroalgae 


Filamentous/turf 


% Cover 


18.1 


2.5 


0.3 


0.1 


<0.1 


<0.1 




Fleshy 


% Cover 


<0.1 


<0.1 


0.1 


<0.1 






Other 


% Cover 


<0.1 


<0.1 








Other 


Anemones 


% Cover 


<0.1 


<0.1 


<0.1 


<0.1 


<0.1 


<0.1 


<0.1 


<0.1 


Other 


% Cover 


4.6 


1.2 


0.5 


0.2 


<0.1 


<0.1 


<0.1 


<0.1 


Tunicates 
(Encrusting) 


% Cover 


2.9 


1.0 


0.3 


0.2 






Tunicates 
(lobate) 


% Cover 


6.3 


1.1 


0.7 


0.1 






Zoanthids 
(benthic dwelling) 


% Cover 


0.4 


<0.1 


<0.1 


<0.1 






Sponge 


Ball 


# Individuals 
% Cover 
Ht (cm) 


0.8 
0.3 
1.7 


0.1 
<0.1 
0.2 


0.4 
0.1 
0.9 


<0.1 
<0.1 
0.2 






Encrusting 


% Cover 


2.4 


0.4 


0.4 


0.2 






Finger 


# Individuals 
% Cover 
Ht (cm) 


0.4 
0.2 
1.3 


<0.1 
<0.1 
0.2 


1.5 
0.3 
2.0 


0.5 
<0.1 
0.4 






Rope 


# Individuals 
% Cover 
Ht (cm) 


0.2 
0.1 
2.1 


<0.1 
<0.1 
0.4 


0.3 
0.1 
2.3 


<0.1 
<0.1 
0.6 






Tube 


# Individuals 
% Cover 
Ht (cm) 


0.9 
0.7 
3.4 


0.1 
0.1 
0.5 


0.8 
0.2 
2.7 


0.2 
<0.1 
0.9 






Vase 


# Individuals 
% Cover 
Ht (cm) 


2.2 
3.5 
12.1 


0.3 
0.4 
1.0 


0.5 
0.4 
2.6 


<0.1 
0.1 
0.5 








dimensions revealed interesting 
differences among the ledge clus- 
ters (Figure 2.7). The three ledge 
clusters were well separated along 
the mean undercut height axis 
such that the tallest ledges gen- 
erally had the highest undercut 
height and the shortest ledges had 
the shortest undercuts. The tallest 
ledges generally also had the larg- 
est average undercut width, but 
there were exceptions to this trend 
(Figure 2.7). 

The mean numbers of burrows in 
flat and rippled sand were 0.92/m 2 
± 0.20 SE and 0.74/m 2 ± 0.23 SE, 
respectively. There was not a sig- 
nificant difference in the number 
of burrows between the two bot- 
tom types (F = 0.3553, df = 35, p 
= 0.55). Mean sand thickness on 
sparse live bottom was 5.4 cm ± 
0.4 SE and ranged from 1.6 cm 
to 14.4 cm (Figure 2.8). The mea- 
sured sand thickness (i.e. within 
quadrats) ranged from cm to 
30 cm, which was the maximum 
depth of sand measured. There 
was a large mode in mean thick- 
ness at 4 cm, with 50% of the sites 
having a mean thickness between 
3.6 and 6.6 cm. 



Q. 
CO 



20" 



12" 



100- 

90- 








Total 


80- 
70- 
60- 




.£ 

\ 




X 2 = 120.8, 
p<0.0001 


50- 
40- 




t 

I 

4. 

1 






30- 
20- 

10- 

o- 


m 






Gorgonians 

%2 = 67.7, 
p<0.0001 



Ledge 



Sparse live 
bottom 



I 
Flat 
sand 



Rippled 
sand 



100- 
90- 




Macroalgae 


80- 
70- 
60- 




k 


X2 = 72.0, 
p<0.0001 


50- 








40- 








30- 




* 




20- 




10- 

o- 




-t 


•■ 



40- 
35- 


■ 




Sponges 


30- 


■ 




X 2 = 109.4, 


25- 






p<0.0001 


20- 




i 






15- 






10- 

5- 

o- 




1 

-J * 
,» 


BB 





100- 






90- 


, 


Other benthic organisms 


80- 
70- 


■ 


X2 = 96.5, 


60- 




p<0.0001 


50- 






40- 
30- 


1 ' 




20- 




10- 

o- 




^ 

\ r 


i. .*. 



Ledge 



Sparse live 
bottom 



Flat 
sand 



Rippled 

sand 



Figure 2.10. Box plots of percent cover of benthic organisms on four bottom substrates 
at GRNMS. Results of nonparametric ANOVAs (Kruskal-Wallis tests) and Dunn's multiple 
comparison tests to determine significant differences among mean ranks are provided (df 
= 3, alpha = 0.05). Solid horizontal lines join groups that are not significantly different from 
each other. 



Ledges in the area of high boat density were sig- 
nificantly larger in area, taller, more undercut, and 
more densely covered by benthic organisms than 
ledges in the low boat density area (Figure 2.9). 
Undercut width was more variable, as exhibited by 
the high standard error, and was not statistically 
different between the two regions. 

Biotic cover 

Summary statistics for biotic composition by bot- 
tom type are displayed in Table 2.6. Multiple com- 
parison tests indicated that cover of coral, mac- 
roalgae, sponges, and other benthic organisms 
was significantly greater on ledges than the other 
bottom types (Figure 2.10). Flat sand and rippled 
sand bottom types were characterized by low per- 
cent cover (0-2%) of benthic organisms at all sites 
(Figure 2.10, Figure 2.11). Percent biotic cover at 
sparse live bottom ranged from 0.7-26.3%, but was 
only greater than 10% at 7 out of 51 sites (Figure 
2.12). On ledge bottom type, percent cover ranged 




Image 11. Undercut ledge densely colonized by tunicates and other 
benthic organisms. 




* — p?~*- s . 



^ \ g**'\ Kilometers 
HT 0.5" -^A<% 




Flat sand 
Rippled sand 



Sparsely colonized live bottom 
Densely colonized live bottom 
GRNMS Boundary 



Percent cover (Flat and Rippled Sand) 



Coral 
Gorgonians 



Macroalgae 
Sponge 
Other 
Bare 



Figure 2.11. Percent cover of biotic cover groups on flat and rippled sand bottom sites. 



from 0.42-100%, with the highest percent cover at ledges in the central and south-central region of GRNMS 
(Figure 2.13). However, percent cover of gorgonians and mean number of gorgonians did not vary significantly 
different between ledge and sparse live bottom sites (Figure 2.10, Figure 2.14). Although a significant difference 
was detected in the height of gorgonians among the four bottom types, no pair-wise tests were significant (Figure 
2.15). In contrast, sponges were significantly more numerous and taller on ledges than on sparse live bottom 
(no sponges were found on either sand bottom type) (Figure 2.15). Shannon-Weiner diversity of biotic types was 
significantly greater at ledge and sparse live bottom than at either sand bottom type (Figure 2.16). 

Cover of corals and gorgonians were generally low (range = 0-18%, mean = 1.35%). Branching coral was the 
most frequently encountered coral type (Figure 2.17), and sea rod/plumes were the most frequently encountered 
gorgonians (Figure 2.18). A high cover of filamentous macroalgae was typical at many of the densely colonized 
ledges (Figure 2.13, Figure 2.19), while several of the northernmost ledges were characterized by high cover of 
sponges, tunicates, and miscellaneous species (including bryozoans, molluscs, barnacles, and other unclassi- 
fied taxa) within the "other" category (Figure 2.20-2.22). Numerous sponge types were observed throughout the 
sanctuary, including encrusting, tube, and vase sponges (Figure 2.20). 





Flat sand 
Rippled sand 



Sparsely colonized live bottom 
Densely colonized live bottom 
GRNMS Boundary 



Percent cover (Sparse Live Bottom) 1 
• ■ 


| Macroalgae 
| Sponge 


| Coral 


Other 






Gorgonians 


1 Bare 



Figure 2.12. Percent cover of biotic cover groups at sparse live bottom sites. 

Cluster analysis of percent cover data for all sites revealed the presence of eight distinct groups of sites (Figure 
2.23). Mean percent cover of the aggregated cover types for each cluster is displayed in Figure 2.24. Cluster 1 
was composed primarily of ledge sites and was characterized by the highest mean percent cover of sponges 
(mean = 23 ± 3% SE). In addition, it has a moderately high (mean = 22 ± 6% SE) mean cover of species within 
the "other" category. Cluster 2 was a larger cluster, again composed of mostly ledges. Total percent cover aver- 
aged just under 30% and was typified by macroalgae, sponges, and other species. Clusters 3-6 contained only 
ledge sites. Clusters 3 and 5 were "outliers" as they contained only one site each and were characterized by 
highest percent cover of coral and gorgonians, respectively. Sites within Cluster 4 typically were characterized 
by the highest percent macroalgal cover. Sites within Cluster 6 were also highly colonized, but these sites were 
dominated by organisms within the "other" category such as tunicates. All of the flat sand and rippled sand sites 
were included in Cluster 7, in addition to several ledges and numerous sparse live bottom sites. Sites within this 
cluster were characterized by low percent cover of all organism types (mean total cover = 1 .8 ± 0.3% SE). Mean 
total cover in Cluster 8, which contained ledge and sparse live bottom sites, was the second lowest among the 
clusters (mean total cover = 8 ± 1% SE) but mean gorgonian cover was higher than many of the other clusters. 
There were no obvious spatial patterns in the distribution of clusters. 












*» 









N 



Flat sand 
Rippled sand 



Sparsely colonized live bottom 
Densely colonized live bottom 
GRNMS Boundary 



Percent cover (Ledge) 1 


| Macroalgae 


H ■ 


| Sponge 


( | Coral 


Other 


Gorgonians 


| Bare 



Figure 2.13. Percent cover of biotic cover groups at ledge sites. 

Cluster analysis of only ledges resulted in six well separated clusters (Figure 2.25). The resulting clusters were 
very similar to those in the first analysis that encompassed all bottom types, with the primary difference being a 
larger cluster containing sites that had been previously dispersed among a few clusters. Similar to the first analy- 
sis, two clusters (4 and 5) contained only one site each that had high percent cover of corals and gorgonians, 
respectively (Figure 2.26). Cluster 1 was characterized by the highest mean percent cover of sponges of all clus- 
ters, as well as moderate-high cover of species within the other category. Cluster 2, containing 52 ledges, was 
characterized by a mean total cover of 28.6 % (± 4.4 SE). Sites in Cluster 3 were densely colonized, particularly 
by macroalgae. Cluster 6 was characterized by the highest mean cover of "other" species, primarily tunicates. 

Abiotic effects on biotic composition 

The effect of ledge height on individual cover types was examined through the use of nonparametric Kruskal- 
Wallis tests. Results revealed significant differences in total biotic cover for all cover types among ledge size 
categories determined from cluster analysis of abiotic variables (Figure 2.27). Median total percent cover was 
97.6%, 75.1%, and 17.7% on tall, medium, and short ledges, respectively. The majority of the ledges classified 
as medium or tall tended to have high overall percent cover. All tall ledges (n=6) had >50% total cover, com- 
pared with 65% of medium ledges (n=26), and 22% of short ledges (n=60). Short ledges had significantly lower 
percent cover than medium or tall ledges, but there was not a significant difference between medium and tall 




CD 

-4— » 

'co 

l_ 
CD 
Q. 

i5 

CO 

■g 
> 

T3 

C 



CD 



20- 
15- 
10- 

5- 
0- 



3 
c 


30- 


c 

CO 




25- 


^ 


20- 



15- 
10- 

5- 
0- 



Gorgonians 
%2 = 65.7, df=3, 
p<0.0001 



■ 



f 



Ledge Sparse live Flat Rippled 

bottom sand sand 



Sponges 
X 2 = 8.8, df=1, 
p=0.003 



^ 



di 



Ledge 



Sparse live 
bottom 



Figure 2.14. Box plots of number of gorgonians and sponges 
on four bottom substrates at GRNMS. Results of nonparametric 
ANOVAs (Kruskal-Wallis tests) and Dunn's multiple comparison 
tests to determine significant differences among mean ranks 
are provided (alpha = 0.05). Sponges were not observed on flat 
or rippled sand bottom types. Solid horizontal lines join groups 
that are not significantly different from each other. 



E 
o 



D5 
'CD 

c 

CO 

CD 



60- 
50- 
40- 
30- 
20- 
10- 
0- 



30- 
25- 
20- 
15- 
10- 

5- 





Gorgonians 
X2 = 9.8, df=3, 
p=0.02 



1^ 



^ 



S T 



a 



i i i i 

Ledge Sparse live Flat Rippled 

bottom sand sand 



Sponges 
X2 = 28.4, df=1, 
p<0.0001 



;* 



4 



m 



.■*■ ■ 



Ledge 



\ 
Sparse live bottom 



Figure 2.15. Box plots of height of gorgonians and sponges on 
four bottom substrates at GRNMS. Results of nonparametric 
ANOVAs (Kruskal-Wallis tests) and Dunn's multiple comparison 
tests to determine significant differences among mean ranks 
are provided (alpha = 0.05). Sponges were not observed on flat 
or rippled sand bottom types. Solid horizontal lines join groups 
that are not significantly different from each other. Although the 
overall test was significant for gorgonian height, no pain/vise 
comparison tests were significant. 



2.5- 
2- 

1.5- 
1- 

0.5- 

o- 



i 

w 

m 
u 
1 

i 

i 
* 


i ■ 
• 

; 

— 
i 



1 



Ledge 



I 



I 



I 
Sparse live 
bottom 



X2 = 88.1, 
p<0.0001 



~~ r 

Flat 
sand 



Rippled 
sand 



Figure 2.16. Box plots of Shannon diversity index (H) by bottom type. Results 
of nonparametric ANOVAs (Kruskal-Wallis tests) and Dunn's multiple compari- 
son tests to determine significant differences among mean ranks are provided 
(df = 3, alpha = 0.05). Solid horizontal lines join groups that are not signifi- 
cantly different from each other. 




f pgt*- s . -■ 

! "T Jf' ' I Kilometers 
*-€ 0.5 **" -„1 -:> 




Flat sand 
Rippled sand 



Sparsely colonized live bottom 
Densely colonized live bottom 
GRNMS Boundary 



Percent cover (Coral) 

6 




Branching 
Cup 

Encrusting 
Other 



Figure 2.17. Percent cover of coral types at ledge sites. The tallest bar in the legend represents 6% cover. 

ledges. Median coral cover was low (<2%) for all ledge types, but was significantly higher on medium than short 
ledges. Percent cover of gorgonians was significantly higher at short ledges when compared with medium or tall 
ledges. Macroalgal cover ranged from 0-76.2% (median = 0.8%) on short ledges, 0-88.4% (median = 21.44%) 
on medium ledges, and 7.6-64.5% (median = 29.9%) on tall ledges. Sponge cover ranged from 0-21 .6% (median 
= 4.35%) on short ledges, 2.84-39.4% (median = 7.9%) on medium ledges, and 8.2-15.2% (median = 10.0%) 
on tall ledges. Cover of other benthic species ranged from 0-33.3% (median = 4.6%) on short ledges, 0.8-86.7% 
(median = 13.2%) on medium ledges, and 21.8-80.3% (median = 38.4%) on tall ledges. Cover of macroalgae, 
sponges, and other benthic organisms was significantly higher at medium or tall ledges when compared with 
short ledges, but there was not a significant difference between medium and tall ledges. 

The number of individual gorgonians was greater at short ledges than at medium or tall ledges (Figure 2.28). 
Short ledges were also characterized by significantly fewer sponges. The number of sponges or gorgonians was 
similar on medium and tall ledges. There was no significant difference in the height of gorgonians between short 
and medium ledges (Figure 2.29); tall ledges were excluded from this analysis because gorgonians were only 
present on one tall ledge. In contrast, sponges were significantly shorter on short ledges compared with the other 
ledge types (Figure 2.29). Shannon diversity did not vary significantly between short, medium, and tall ledges (F 
= 0.24, df = 91, p = 0.79). 




'' ■ ^tX^':^ ' .^ — < - r ^5' j^P*- 5 

*£p| ,J I g*'\ Kilometers 
i -*f 0.5 ^< ! 



-V 



*>.. 




ff} ;.rr ^V'-- - 



_^n 



v. 







N 



Flat sand 
Rippled sand 



Sparsely colonized live bottom 
Densely colonized live bottom 
GRNMS Boundary 



Percent cover (Gorgonians) 



J]« 



Sea whip 

Sea fan 

Sea rod/plume 



Figure 2.18. Percent cover of gorgonian types at ledge sites. The tallest bar in the legend represents 10% cover. 

2.5 DISCUSSION 

GRNMS is composed of four main bottom types that have distinct physical and biological characteristics. In sup- 
port of ongoing management of sanctuary resources, the overall goal of this component of the characterization 
was to quantify abiotic features and cover of benthic species. This study builds on previous work of benthic habi- 
tats in GRNMS, which includes benthic habitat mapping (Kendall et al. 2005), a guide to invertebrates (Gleason 
et al.), macroalgae (Searles 1988), a characterization of soft-bottom macrobenthos (Hyland et al. 2006) and 
several studies that surveyed individual species or communities associated with live bottom within GRNMS 
(Hopkinson et al. 1 991 ; Ruzicka 2005; Wagner 2006). The primary difference between this characterization and 
prior surveys is that the present study encompassed all habitat types and quantified both abiotic characteristics 
and epibenthic communities at a large number of site locations throughout the sanctuary. 

Our first objective was to assess the accuracy of the habitat maps by Kendall et al. (2005). Overall map ac- 
curacy was excellent at 93.3% as measured in the present study. Kendall et al. (2005) previously estimated a 
similarly high level of overall thematic accuracy at 94.8% correct although quite different methods were used. 
While the present assessment used random stratified points and diver based assessments, Kendall et al. (2005) 
used randomly placed video transects and spatial statistics. Kendall et al. (2005) were limited to assessing sand 
(both rippled and flat combined) and sparse live bottom due to the nature of the transect and video based data. 
Quantitative assessment of the accuracy of the ledge category was not possible. Considering only sparse live 





Flat sand 
Rippled sand 



Sparsely colonized live bottom 
Densely colonized live bottom 
GRNMS Boundary 



Percent cover (Macroalgae) 



dl 



50 



Filamentous 

Fleshy 

Other 



Figure 2.19. Percent cover of macroalgae types at ledge sites. The tallest bar in the legend represents 50% cover. 

bottom and sand, similarly high values were found by Kendall et al. (2005) and in the present study despite quite 
different methods. For sparse live bottom, Kendall et al. (2005) found user's and producer's accuracies of 90.9% 
and 93.0% respectively, compared to 100% and 86.3% in the present study. For sand, Kendall et al. (2005) 
found user's and producer's accuracies of 96.7% and 95.7% respectively whereas in the present study we found 
values of 1 00% and 94.4% (when results from rippled and flat sand were combined). In the present study, user's 
and producer's accuracies of the ledge category were 91.1% and 100%, respectively. The combined findings of 
these two studies demonstrate a robust and complete accuracy assessment of all the bottom types at GRNMS. 

All nine errors in the ledge category of the present assessment occurred in the August 2004 sampling period. At 
two sites which were surrounded by large areas of sand on all sides, divers' notes indicated the presence of a 
deflection in bathymetry that was sparsely colonized with sessile benthic invertebrates. These ledges were prob- 
ably better defined in 2001 when the sonar data were collected, but were now in the process of being covered 
by shifting sands. Further along the length of these features, the ledge may have been better defined. However, 
because a ledge was not readily observable at either of these two sites, the divers noted that they moved off a 
short distance and did a survey over rippled sand in order to not waste the dives. At two other sites identified 
as sparse live bottom, the divers noted that ledges were nearby but that the survey was conducted on sparse 
live bottom. No special notes were made for five other ledge sites identified as sparse live bottom. After this first 
sampling period it became apparent that some small ledges could be missed by divers searching for them un- 





Flat sand 



Rippled sand 



Sparsely colonized live bottom Percent cover (Sponge) 
Densely colonized live bottom 
GRNMS Boundary 



10 




Ball 




Encrusting 




Finger 



Rope 
Tube 
Vase 



Figure 2.20. Percent cover of sponge types at ledge sites. The tallest bar in the legend represents 10% cover. 



derwater due to the often limited visibility at GRNMS. For later sampling trips, the starting point of each randomly 
selected ledge was moved based on sonar images to a segment of the ledge that was likely to be more easily 
located underwater. Once on a ledge it can be followed more easily through areas even where it is less promi- 
nent. Indeed no other ledges were missed in the subsequent two sampling periods once this modification to site 
selection was made. Some combination of actual map errors, ledges being covered by sand, and diver error is 
probably responsible for the large number of discrepancies between map and diver opinion in the August 2004 
sampling period. 

All three remaining errors were in the sand category and occurred during the August 2005 sampling period. This 
most recent of the sampling periods was four years and two months after the sonar data used to develop the 
benthic maps were collected in June 2001. This time gap would allow time for bottom altering forces such as 
bioturbation to rework the surficial sediments (Sisson et al. 2002) and for localized water movements to create 
or remove sand ripples. Indeed, the presence of many echinoderms reworking the surface was noted at one site 
that was mapped based as rippled sand (2001 data) but was later identified as flat sand in 2005. 

In situ ledge height estimates confirmed sonar derived estimated of ledge heights. Ledge height estimated from 
GIS analysis was positively correlated with maximum ledge height measured in situ although GIS height was 
always higher than in situ estimates. There are two possible explanations for this discrepancy. First, in situ ledge 





-*■• afe*-s 



I I iff"' I Kilometers 

*"& 0.5 ••=„ 1 <5 



1= 



■>■• -t 



^-p 




•-r-i 



5/ 



N 

*A E 



Flat sand 
Rippled sand 



Sparsely colonized live bottom 
Densely colonized live bottom P 
GRNMS Boundary 



Percent cover (Other-Tunicates) 

50 



Tunicates (Encrusting 
Tunicates (Lobate) 



Figure 2.21. Percent cover of tunicates at ledge sites. The tallest bar in the legend represents 50% cover. 

height was only measured at five points along a 25 m transect. Many ledges were longer than this and it is quite 
possible that the tallest part of the ledge simply wasn't measured. Second, differences in height estimates may 
be a product of the way GIS height was determined. Height was calculated as the difference between the deep- 
est and shallowest bathymetry values. In the case of a broadly sloping ledge, it is likely that these points would 
be located at opposite ends of the ledge. In such a case, height measured by a diver at any location along the 
ledge would be lower than the maximum GIS height. 

Biological and physical processes work to continually shape the sand and hardbottom features. The sand and 
shell rubble observed on sparse live bottom sites may have been deposited through sand movement from 
nearby sandy areas or from weathering of hardbottom. Fine sediments were not observed, which is consistent 
with previous records of sediment distribution on the mid-continental shelf in the South Atlantic Bight (Milliman et 
al. 1972; Riggs et al. 1996; Hyland et al. 2006). Large storms and seasonal storm patterns can cause sediments 
to shift, which may alter benthic communities or result in an import/export of sediments to the system (Riggs et 
al. 1998). This constant shifting of sediments likely prevents the flat hardbottom from becoming more densely 
colonized by epibenthic fauna. The sediment layer covering the bottom was usually several centimeters thick 
(up to 18 cm) which would prevent larvae from settling, or bury recent recruits. The most common cover type on 
sparse live bottom was sea rods/plumes, which are often quite tall (mean height = 19.4 cm), making them less 
vulnerable to burial. Furthermore, these gorgonians may have colonized the sparse live bottom areas when such 





Figure 2.22. Percent cover of "other" benthic cover types at ledge sites. The tallest bar in the legend represents 50% cover. 

areas were uncovered by sand or may have growth rates that exceed the rate of sand deposition on hardbottom. 
At some sites, divers cleared away sand that covered sparse live bottom and noted living sponges and ascidians 
that may have recently been buried by sand. In addition, the mean number of sponges on sparse live bottom 
was not significantly different from ledges, but sponges were smaller and covered a smaller percentage of the 
substrate on sparse live bottom. Sponge morphology also differs between different bottom types. In a recent 
study at GRNMS and nearby J Reef, Ruzika (2005) documented distinct sponge communities at ledge "scarps" 
(ledges in the present study) and "plateaus" (similar to sparse live bottom in the present study). The majority of 
species occurring on the scarps were amorphous or encrusting species, while the plateaus were characterized 
by branching, pendunculate, or digitate sponges (Ruzika 2005). 



Numerous cover types were observed on ledges, including macroalgae, sponges, tunicates, coral, and gorgoni- 
ans. High diversity of macrofauna in Gray's Reef and other stations in the inner and mid-shelf was also observed 
by Wenner et al. (1983). Although that survey was conducted using dredge and trawl methods, the taxonomic 
groups of major importance included sponges, bryozoans, corals, anemones, tunicates, and echinoderms. Hop- 
kinson et al. (1991) documented similar taxa in GRNMS in association with a survey of community metabolism 
at an east-central site along the northern rim of ledges and hardbottom. Dominant morphology types included 
sponges, corals, and miscellaneous species (bryozoans, hydroids, ascidians, and mussels). Hopkinson et al. 
(1991) also found higher coral cover than was observed at any site in the present study; however, the one area 




I sampled may have been a hot spot for coral cover. Thus, it may be inappropriate to compare our average esti- 
mates of coral cover to that reported by Hopkinson et al. (1991). In addition, in the previous study, mean mac- 
roalgal cover did not exceed 9% at low, medium, and high density areas (Hopkinson et al. 1991). 

Linking biological community structure to the environment is a major goal of ecology but often is difficult to as- 
sess (Clarke and Ainsworth 1993). Geological differences in substrate type or morphological complexity have 
been linked to community structure, but the patterns are not always universal (Davis et al. 2003; Beaman et 



Cluster 1: 6 L, 1 SLB 
High sponge cover 

Cluster 6: 6 L 

Highest mean total cover 

Dominated by other organisms 

Cluster 3: 1 L 
Highest % coral cover 

Cluster 5: 1 L 

Highest % gorgonian cover 

Cluster 4: 27 L 

Highest % macroalgal cover 

Cluster 2: 27 L, 2 SLB 
Low -moderate cover, composed of 
macroalgae , sponges, and other 
benthic organisms 



Cluster 8: 13 L, 16 SLB 
Low overall cover 



Cluster 7: 21 FS, 15 RS, 

11 L, 32 SLB 

Very low overall cover 




■ w " w " v " w " vwmwi 



Figure 2.23. Dendrogram from cluster analysis of all sites based on percent cover for aggregated cover types (corals, gorgo- 
nians, macroalgae, sponge, other). The scree plot indicates that there are eight well separated clusters. Sites are color coded 
by bottom type (dark blue=ledge, light blue=sparse live bottom, yellow=flat sand, orange=rippled sand). 



al. 2005). The majority of ledges 
surveyed in GRNMS were short 
with little or no undercut, while 
ledges classified as medium or 
tall exhibited varying amounts 
of undercut. Although undercut 
dimensions were positively cor- 
related with ledge height, there 
were exceptions to this trend 
(Figure 2.7). For example, the 
tallest ledge surveyed (mean 
height = 170 cm) had relatively 
little undercut (mean undercut 
height = 33.8 cm, mean under- 
cut width = 25 cm). Tall ledges 
with small undercuts may result 
from physical and bioerosional 
processes that may weaken por- 
tions of the ledge overhangs over 
time, eventually causing them to 
fall off and form rock rubble in 
the ledge openings (Riggs et al. 
1996; Riggs et al. 1998). 

In general, medium and tall ledg- 
es did not differ in total biotic cover 
or the cover of individual morpho- 
logical groups. It is possible that 
height may only be important to a 
certain threshold and that ledges 
above a certain height are sim- 
ply less likely to be routinely bur- 
ied and unburied by sediments. 
There was also a large degree 
of variability in cover on short 
ledges. Despite the low median 
cover, a few short ledges had to- 
tal cover exceeding 50%. How- 
ever, excluding gorgonians and 
coral, cover of other groups was 
significantly less on short ledges. 
Compared to tall ledges, low re- 
lief ledges would likely be more 
subject to burial by shifting sedi- 
ments, which could inhibit coloni- 
zation by organisms. Several of 
these short ledges with low cov- 
er are located in the southeast 
corner of the sanctuary and are 
otherwise surrounded by sand. 
This would be a good area to 
investigate sand migration rates 
and the associated impacts on 
ledges. Gorgonians and several 
sponge types (finger, rope, and 



> 
O 
O 
-t— i 

c 
<u 
o 

i_ 

<u 

c 

TO 





ft 



□ Other 

■ Sponge 

■ Macroalgae 

□ Gorgonians 

■ Coral 



Cluster 

Figure 2.24. Mean percent cover of benthic organisms (±SEM) by clusters determined 
from hierarchal cluster analysis in Figure 2.23. 



Cluster 1 : 6 L 

Highest mean % cover of 

sponges 

Cluster 6: 8 L 

Highest mean % cover of 

"other " benthic species 



Cluster 3: 24 L 

Highest mean % cover of 
macroalgae 



Cluster 2: 52 L 

Low mean total cover 



Cluster 4: 1 L 

Highest % coral 



Cluster 5: 1 L 
Highest % gorgonian 




Figure 2.25. Dendrogram for cluster analysis of the 92 ledge sites based on percent 
cover for aggregated cover types (corals, gorgonians, macroalgae, sponge, other). The 
scree plot indicates that there are six well separated clusters. The numbers representing 
each site denote total mean percent cover. 




tube) appear to be less limited than 
other invertebrates and were found 
on sparse live bottom or short ledges 
with similar or greater frequency than 
on medium or tall ledges. While many 
of the ledges located in the southeast 
region were characterized by low 
overall cover, gorgonians and a low 
density of sponges were nearly al- 
ways present in this area. 

Conversely, taller ledges and slopes 
would be less susceptible to sand 
burial and can support shorter colo- 
nies. Total cover increased with ledge 
height, but examination of spatial pat- 
terns and cluster analyses by biotic 
type indicate that a diversity of benthic 
community combinations occur re- 
gardless of ledge height. The 
most densely colonized ledge 
sites were generally situated 
in the central region of the 
sanctuary, particularly sites 
located among the northern 
rim of ledges and a group of 
ledges to the south (Figure 
2.13). Highest cover of coral, 
macroalgae, sponges, and tu- 
nicates tended to be located 
in these regions as well (Fig- 
ures 2.13, 2.17-2.21). 



> 
o 
o 

c 

o 

<D 
Q_ 

c 

TO 
<D 



100 - 
80 - 




-^-i 


-fr 




60 - 
40 - 


rfi 


-i- 












20 - 

n - 


* a 






u 








-E- 





□ Other 

■ Sponge 

■ Macroalgae 

□ Gorgonians 

■ Coral 



12 3 4 5 6 

Cluster 

Figure 2.26. Mean percent cover (±SEM) by cluster determined from hierarchal cluster- 
ing of ledges (Figure 2.24). 



The reasons for these spatial 
patterns are not clear. Many of 
the densely colonized ledges 
were also tall in height, but this 
was not always the case. Why 
some ledges were dominated 
by macroalgae, and others of 
similar height by tunicates is 
unknown. It is likely that other 
factors not considered in this 
study, such as small scale 
rugosity or complexity, ocean 
currents, other environmen- 
tal variables, and biological 
interactions (e.g., differential 
grazing or settlement pat- 
terns) work in concert to influ- 
ence spatial distribution and 
community structure. Osman 
(1977) notes five major fac- 
tors important to the develop- 
ment of an epifaunal benthic 




100- 




90- 
80- 
70- 


X 2 = 10.8, 
p=0.0S46 










60- 


1 

T 










50- 
40- 
30- 


f 




■ 




% 


20 








i 






■|- 




10 




L 

























CD 
> 
O 


75 
P 



c 

CD 
O 

!_ 

CD 
Q. 

C 
00 
CD 



20- 



16- 



12- 



Coral 

X 2 = 16.4, p=0.0003 



20- 
15- 

10- 
5- 

o- 



Gorgonians 
X 2 = 24.4, 
p<0.0001 



40- 
35- 
30- 
25- 
20- 
15- 
10- 
5- 

o- 



100- 

90- 
80- 
70- 
60- 
50- 
40- 
30- 
20- 
10- 

o- 





■ 


Sponges 

X 2 =18.3, 
p=0.0001 




T 

■* 


E£ 







Other benthic organisms 
X 2 = 30.9, — ~ 

p<0.0001 



Short 



Medium 



Tall 



Short 



Medium 



Tall 



Figure 2.27. Box plots of percent cover of benthic organisms on three ledge groups determined 
by cluster analysis. Results of nonparametric ANOVAs (Kruskal-Wallis tests) and Dunn's multiple 
comparison tests to determine significant differences among mean ranks are provided (df = 2, al- 
pha = 0.05). Solid horizontal lines join groups that are not significantly different from each other. 




CD 
■ i— ' 

CO 

1— 
CD 

Q_ 

iG 

CD 

13 

-g 
'> 

C 



CD 

E 

13 
C 

CO 
CD 



6- 
5- 


Gorgonians 

X 2 = 21.6, p<0. 0001 


4- 






3- 




1 




2- 




f 








1- 




# 










o- 


* 


-^-=- 


-*- 



3.5" 

3- 


Sponges 

X 2 = 15.1, p=0. 0005 ■ 




2.5- 


-M— 






2- 


—r~ 








1.5- 


T n- 








1- 

0.5- 




1\ 
^ ■ 

4r 




1 


\ 




j 




U 









I 

Short 



I 
Medium 



I 
Tall 



Figure 2.28. Box plots of mean number of gorgonians 
and sponges on three ledge groups determined by clus- 
ter analysis of abiotic ledge variables. Results of non- 
parametric ANOVAs (Kruskal-Wallis tests) and Dunn's 
multiple comparison tests to determine significant differ- 
ences among mean ranks are provided (df = 2, alpha 
= 0.05). Solid horizontal lines join groups that are not 
significantly different from each other. 



E 
o 



cu 

c 
CO 

cu 



60- 
50- 




■ 


Gorgonis 

X 2 =1.6,c 


ns 
f=1 


P= 


121 


40- 


. 


■■ 






30- 








20- 




.' 






- 


_i- 




■ ■ 








10- 






.■- 


- 


■■ 




" t. 














u- 











Short 



Medium 



30- 


Sponges 
X 2 = 9.2, df= 


=2p= 










25- 


=0.01 




" 




20- 


— f— 








15- 


f 




— f— 






10- 
5- 

o- 




.V 








•i 


i — 




* 
J 





Short 



Medium 



Tall 



Figure 2.29. Box plots of mean height of gorgonians and 
sponges on three ledge groups determined by cluster analysis. 
Results of nonparametric ANOVAs (Kruskal-Wallis tests) and 
Dunn's multiple comparison tests to determine significant dif- 
ferences among mean ranks are provided (alpha = 0.05). Only 
short and medium ledges were considered for the analysis of 
gorgonian height. Solid horizontal lines join groups that are not 
significantly different from each other. 



community and species distribution, including larval selection of settlement location, seasonal fluctuation in the 
climate and larval abundance, biological interactions (predation, intra and inter-specific competition), substrate 
size, and physical disturbance. In particular, the structural complexity of ledges is difficult to quantify, but it may 
contribute greatly to species diversity and variability between different patches. For example, within a small area 
(e.g. < 100 m 2 ) numerous microhabitats may exist (e.g., crevices, overhangs, bare rock, sand patches) that may 
allow a large number of species to co-exist due to differential larval settlement and survival patterns (Wenner et 
al. 1983). 

Although flat and rippled sand bottom types were largely devoid of epibenthic invertebrates and macroalgae 
that settle on hardbottom, this does not mean that this habitat is not important ecologically. Numerous burrows 
indicate the presence of benthic infauna. Indeed, Hyland et al. (2006) documented 349 different infaunal taxa, 
including polychaetes, crustaceans, echinoderms, and mollusks from sediment grabs in GRNMS. Patterns in 
species composition and abundance varied among bottom types. The diversity and densities of macroinfauna 
were higher in rippled sand than flat sand, and were also high in sediments close to or overlying live bottom (Hy- 
land et al. 2006). Macroinfauna serve as important prey items for numerous fish in GRNMS, including grunts and 
Pareques species. Fish such as these are closely associated with ledges but often forage over adjacent sand 
bottoms, resulting in a "halo effect" of decreased infaunal abundance closer to reef ledges (Posey and Ambrose 
1994). 

Although this study was not designed to detect seasonal patterns, invertebrates and particularly macroalgae are 
prone to both seasonal and year-to-year variability (Peckol and Searles 1984). Seasonal and inter-annual differ- 
ences can be attributed to multiple sources, including fluctuating environmental conditions (Peckol and Searles 
1984), storm events (Renaud et al. 1997), and success in larval settlement and growth (Osman 1977). GRNMS 
lies at the boundary between the inner and middle continental shelf and tropical and temperate water masses, 
and hence experiences seasonal fluctuations in temperature, salinity, and water clarity (NOAA 2006). Coastal 




circulation patterns in the South Atlantic Bight are 
also prone to seasonal variation (Bumpus 1955). 
Mean water temperature at the GRNMS data buoy 
was 30°C in August 2005, two degrees higher than 
the previous August. Mean water temperature in 
May 2005 was much cooler at 23.4°C (data ob- 
tained from the NOAA Station 41008 buoy located 
in GRNMS, http://www.ndbc.noaa.gov/, accessed 
Sept. 30, 2006). Although mean cover for most 
groups was similar across the three sampling peri- 
ods, there is some indication of temporal patterns 
for macroalgae. At Gray's Reef, macroalgae gen- 
erally reaches peak abundance in July and August 
before dying back in the fall and winter (Searles 
1988). In the present study, mean macroalgal cov- 
er on ledges varied from 0.6% in August 2004 to 
1.3% in May 2005 to 11.6% in August 2005. No 
ledges sampled in August 2004 or May 2005 had a 
mean macroalgal coverage exceeding 25%. Con- 
versely, 16 ledge sites sampled in August 2005 

were covered by macroalgae in excess of this amount. Results from cluster analysis also indicated some tempo- 
ral differences, as all but one site located in the "macroalgae" cluster was surveyed in August 2005. Due in part 
to this macroalgal bloom, the seven most densely colonized ledges (99-100% total cover) were documented in 
August 2005. The causes for the large difference in macroalgae in the two successive summers are unknown, 
but in addition to interannual variation in environmental conditions and storm events, macroalgal growth is also 
sensitive to variability in nutrient levels and grazing pressure (Miller and Hay 1996). Similarly, Peckol and Searles 
(1984) detected strong interannual differences in macroalgal cover on North Carolina reefs, but less so for in- 
vertebrates. In addition, it cannot be ruled out that observed differences in macroalgal cover in the present study 
were partially attributed to the random sampling design as the same sites were not characterized across all three 
surveys. More work is needed to accurately assess temporal patterns in macroalgae and epifaunal cover and 
responsible control mechanisms. 




Image 12. Flamingo tongue and gorgonian. 



Temperate reefs such as those in GRNMS differ from coral reefs in other National Marine Sanctuaries (Florida 
Keys, Flower Garden Banks, NWHI) in numerous ways, including geologic origin (Harding and Henry 1990) and 
dominant biota (Miller and Hay 1996). Unlike tropical reefs, temperate reefs consist of pre-existing, submerged 
rocky outcrops that are colonized by epibenthic organisms (Harding and Henry 1990). Corals are less common 
on temperate reefs and tend to form smaller colonies then in tropical regions (Miller and Hay 1996). The latitudi- 
nal limits of coral are thought to be attributed not only to lower temperatures but also increased competition with 
macroalgae, which are favored in higher nutrient waters (Johannes et al. 1983). Oculina arbuscula, the primary 
coral species in GRNMS, ranges from the Carolinas to Florida (Humann 1993) and has a wide temperature 
tolerance, although highest growth occurs in warm water under high light conditions (Miller 1995). The distribu- 
tion of Oculina on temperate reefs in North Carolina is limited by macroalgae through both direct competition 
and indirectly by restricting the coral to deeper, lower lit environments where Oculina growth is not as favorable 
(Miller and Hay 1 996). Recent work by Wagner (2006) on the population genetics of O. arbuscula in GRNMS and 
surrounding hardbottoms provide evidence for local recruitment, perhaps due to the nature of the patchy reef 
environment. In the present study, coral was commonly observed at 75% of all ledge sites, however, it generally 
contributed a small percentage to total percent cover. 

In contrast, sponges represent an important component of the benthic community in GRNMS, accounting for as 
high as 39% cover. Usually multiple morphological types and species were present in a single quadrat at an in- 
dividual ledge. Although less studied, sponges often exceed corals and algae in terms of diversity on coral reefs 
(Diaz and Rutzler2001), and some species may compete with coral for space (Aerts 1998). Compared to tropical 
reefs, temperate SAB reefs appear to have lower species diversity, but higher density of species and individuals, 
particularly for encrusting species (Ruzika 2005). 



The South Atlantic Bight is occasionally affected by tropical cyclones and strong winter storms, which can result 
in strong current overflow and turbulence over ledge outcrops, particularly near the lip of the ledge (Peckol and 
Searles 1984). Although Peckol and Searles (1984) found reduced colonization by perennial macroalgae in this 
environment compared to several meters back from the face of the ledge, sessile invertebrates appeared to be 
less restricted and colonized this region heavily In addition, surveys of North Carolina reefs following Hurricane 
Diana found little damage to the benthic invertebrate communities, which suggests that benthic communities 
are resilient to the impacts of strong storms and bottom currents (Kirby-Smith and Ustach 1986; Vaughan et 
al. 1987). Furthermore, although storms can negatively affect algae and invertebrates through dislodgement or 
scouring, they may also create favorable conditions for settlement by exposing hardbottom that had previously 
been covered by sediments (Renaud et al. 1 997). Although insufficient data is available to investigate this further, 
differential storm patterns between years is one possible factor that may have contributed to the differences in 
cover of macroalgae in 2004 and 2005. 

Concerns were raised about potential human impacts on the sanctuary resources in the recently updated GRNMS 
management plan (NOAA2006). Compared to other hardbottom habitats, regulations afford the sanctuary pro- 
tection from trawling and dredging, which have been shown to damage sponges, gorgonians and corals (Van 
Dolah et al. 1987). However, recreational activities can also negatively impact benthic fauna. For example, our 
surveys found fishing line entangled in oculinid coral (Chapter 3). Ledges within the area of high boat density 
were on average larger in area, taller, and had a higher percent cover than in the area where less boat activity 
was observed. This is not surprising as fishermen are more likely to detect larger, taller ledges on their depth 
finder (personal communication), and these ledges are more likely to harbor larger fish densities (Chapter 4). 
However, this finding is significant to management because it indicates that areas with the highest amount of live 
bottom may be disproportionately more vulnerable to human impacts (e.g., anchoring, derelict fishing gear). 

2.6 RECOMMENDATIONS FOR MANAGEMENT AND MONITORING 

Monitoring sanctuary resources on an annual basis is crucial to understand year-to-year variations in abundance 
and presence or absence of epibenthic flora and fauna, and to assess any changes in condition of biological 
communities over time. This work provides a baseline assessment and a foundation for long-term monitoring of 
the benthic community in GRNMS. Surveys of the benthos can be conducted in conjunction with the transect 
surveys for fish using the procedures outlined in the field methods of this document. We recommend conducting 
the annual survey in the summer of each year, as this is when annual species such as macroalgae are likely to 
be highest. Should additional resources remain, a survey could also be conducted during another time of the 
year to address seasonal variations. Ledges should receive a majority, if not all, of the effort due to the high 
abundance and diversity of sessile flora and fauna and associated fish (Chapter 4). Ledges of all sizes and 
heights should be surveyed to further characterize 
the relationship of invertebrate communities with 
their environment. Although diversity was also high 
on sparse live bottom, cover of most biota, with the 
exception of gorgonians, was significantly lower on 
this bottom type. 

Information from this baseline survey can be used 

to adjust field methods in the future. For example, 

due to the prevalence of tunicates on ledges, tu- 

nicates could be given their own category (with 

subcategories lobate and encrusting) rather than 

subcategories under "other" benthic cover types. 

Other potential categories/subcategories that were 

not included on data sheets, but were occasionally 

noted, include barnacles and particularly bryozo- 

ans. One factor that was not measured at ledge 

sites was the thickness of overlying sediment, but 

as this is important to community development and 

Image 13. Undercut ledge densely colonized by coral, sponges, and 
other benthic organisms. 





I maintenance, quantifying this variable may shed additional light on the dynamics of benthic community pat- 
terns. 

The field survey method applied in this study is advantageous in that it allows researchers to survey a large num- 
ber of sites within a short period of time. To date, this was the most spatially comprehensive characterization of 
benthic communities at GRNMS and yielded a substantial amount of information on the abundance and distribu- 
tion of benthic invertebrates within GRNMS and their association with major bottom types. However, the method 
is not without drawbacks. For example, due to time constraints, invertebrates and macroalgae were identified 
by morphology rather than species. Additional surveys could focus on a subset of sites and specific cover types 
(e.g. sponges, corals) to identify individuals to a lower taxonomic level. In addition, studies pertaining to recruit- 
ment, settlement, and population dynamics of invertebrate species in GRNMS (e.g., Wagner 2006) should be 
continued. The invertebrate species database (http://www.bio.georgiasouthern.edu/GR-inverts/, Gleason et al.) 
is a valuable public-accessible resource that should continue to be updated over time. 




REFERENCES 

Aerts, LAM. 1998. Sponge/coral interactions in Caribbean reefs: Analysis of overgrowth patterns in relation to species iden- 
tity and cover. Marine Ecology Progress Series 175:241-249. 

Beaman, RJ, JJ Daniell, and PT Harris. 2005. Geology-benthos relationships on a temperate rocky bank, eastern Bass 
Strait, Australia. Marine and Freshwater Research 56(7): 943-958. 

Bumpus, DR 1955. The circulation over the continental shelf south of Cape Hatteras. Transactions of the American Geo- 
physical Union 36(4): 601-611. 

Clarke, KR, and M Ainsworth. 1 993. A method of linking multivariate community structure to environmental variables. Marine 
Ecology Progress Series 92: 205-219. 

Congalton, RG, and K Green. 1999. Assessing the accuracy of remotely sensed data: principles and practices. CRC Press 
Inc., Lewis Publishers, Boca Raton, FL. 

Davis, AR, SK Fyfe, X Turon, and MJ Uriz. 2003. Size matters sometimes: wall height and the structure of subtidal benthic 
invertebrate assemblages in south-eastern Australia and Mediterranean Spain. Journal of Biogeography 30(12): 1797- 
1807. 

Diaz, MC, and K Rutzler. 2001. Sponges: An essential component of Caribbean coral reefs. Bulletin of Marine Science 
69(2): 535-546. 

Gleason, DF, AW Harvey, and SP Vives. A Guide to Benthic Invertebrates and Cryptic Fishes of Gray's Reef. Georgia South- 
ern University. Statesboro GA. http://www.bio.georgiasouthern.edu/GR-inverts/. 

Hanson, RB, KR Tenore, S Bishop, C Chamberlain, MM Pamatmat, and J Tietjen. 1 981 . Benthic enrichment in the Georgia 
Bight related to Gulf Stream intrusions and estuarine outwelling. Journal of Marine Research 39(3): 417-441. 

Harding, JL, and VJ Henry. 1990. Geological History of Gray's Reef National Marine Sanctuary, http://graysreef.noaa.gov/ 
geology.html. 

Henry, VJ, and RT Giles. 1979. Distribution and occurrence of reefs and hardgrounds in the Georgia Bight. USGS Office of 
Marine Geology, 55 Woods Hole, Massachusetts. 

Hopkinson, CS, RD Fallon, BO Jansson, and JP Schubauer. 1991. Community metabolism and nutrient cycling at Grays 
Reef, a hard bottom habitat in the Georgia Bight. Marine Ecology Progress Series 73: 105-120. 

Humann, P. 1993. Reef Coral Identification. Paramount Miller Graphics, Inc., Jacksonville, FL. 

Hyland, J, C Cooksey, WL Balthis, M Fulton, D Bearden, G McFall, and M Kendall. 2006. The soft-bottom macrobenthos 
of Gray's Reef National Marine Sanctuary and nearby shelf waters off the coast of Georgia, USA. Journal of Experimental 
Marine Biology and Ecology: A Tribute to Richard M. Warwick 330(1): 307-326. 

Johannes, RE, WJ Weibe, CJ Crossland, DW Rimmer, and SV Smith. 1983. Latitudinal limits of coral reef growth. Marine 
Ecology Progress Series 11:105-111. 

Kendall, MS, and KA Eschelbach. 2006. Boundary options for a research area within Gray's Reef National Marine Sanctu- 
ary. NOAA Technical Memorandum NOS NCCOS 31, Silver Spring, MD. 

Kendall, MS, OP Jensen, C Alexander, D Field, G McFall, R Bohne, and ME Monaco. 2005. Benthic mapping using sonar, 
video transects, and an innovative approach to accuracy assessment: A characterization of bottom features in the Georgia 
Bight. Journal of Coastal Research 21(6): 1154-1165. 

Kennedy, J. 1993. Gray's Reef National Marine Sanctuary: after a decade in the sanctuary program, it still has secrets to 
share. Mariner's Weather Log 37:2a-8a. 

Kirby-Smith, WW, and J Ustach. 1986. Resistance to Hurricane Disturbance of an Epifaunal Community on the Continental- 
Shelf Off North-Carolina. Estuarine Coastal and Shelf Science 23(4): 433-442. 

Miller, MW. 1995. Growth of a Temperate Coral - Effects of Temperature, Light, Depth, and Heterotrophy. Marine Ecology- 
Progress Series 122: 217-225. 




Miller, MW, and ME Hay. 1996. Coral-seaweed-grazer-nutrient interactions on temperate reefs. Ecological Monographs 
66(3): 323-344. 

Milliman, JD, OH Pilkey, and DA Ross. 1 972. Sediments of the continental margin off the eastern United States. Geological 
Society of America Bulletin 83:1315-1334. 

NOAA. 2006. Gray's Reef National Marine Sanctuary Final Management Plan/Final Environmental Impact Statement. 
NOAA NOS NMSP, Savannah, GA. 

Osman, RW. 1977. The establishment and development of a marine epifaunal community. Ecological Monographs 47:37- 
63. 

Parker, RO, DR Colby, and TD Willis. 1983. Estimated amount of reef habitat on a portion of the United States South Atlantic 
and Gulf of Mexico Continental Shelf. Bulletin of Marine Science 33(4): 935-940. 

Peckol, P, and RB Searles. 1984. Temporal and spatial patterns of growth and survival of invertebrate and algal populations 
of a North Carolina continental shelf community. Estuarine Coastal and Shelf Science 18(2): 133-143. 

Posey, MH, and WG Ambrose. 1994. Effects of proximity to an offshore hard-bottom reef on infaunal abundances. Marine 
Biology 118(4): 745-753. 

Renaud, PE, SR Riggs, WG Ambrose, K Schmid, and SW Snyder. 1 997. Biological-geological interactions: Storm effects on 
macroalgal communities mediated by sediment characteristics and distribution. Continental Shelf Research 1 7(1 ): 37-56. 

Riggs, SR, WG Ambrose, JW Cook, and SW Snyder. 1998. Sediment production on sediment-starved continental mar- 
gins: The interrelationship between hardbottoms, sedimentological and benthic community processes, and storm dynamics. 
Journal of Sedimentary Research 68(1): 155-168. 

Riggs, SR, SW Snyder, AC Hine, and DL Mearns. 1996. Hardbottom morphology and relationship to the geologic frame- 
work: Mid-Atlantic continental shelf. Journal of Sedimentary Research 66(4): 830-846. 

Ruzika, R.R. 2005. Sponge community structure and anti-predator defenses on temperate reefs of the South Atlantic Bight. 
Master of Science, Georgia Southern University, Statesboro, GA. 

Searles, RB. 1988. An Illustrated Field Guide to the Seaweeds of Gray's Reef National Marine Sanctuary. NOAA technical 
memorandum NOS MEMD 22, Silver Spring, MD. 

Sedberry, GR, and RF Van Dolah. 1984. Demersal Fish Assemblages Associated with Hard Bottom Habitat in the South- 
Atlantic Bight of the USA. Environmental Biology of Fishes 11(4): 241-258. 

Sisson, JD, J Shimeta, CAZimmer, and P Traykovski. 2002. Mapping epibenthic assemblages and their relations to sedi- 
mentary features in shallow-water, high-energy environments. Continental Shelf Research 22(4): 565-583. 

Sokal, RR., & Rohlf, FJ. 1995. Biometry: The Principles and Practice of Statistics In Biological Research. (3rd ed.). Stony 
Brook, NY, USA.: W.H. Freeman and Company. 

Van Dolah, RF, PP Maier, GR Sedberry, and CA Barans. 1994. Distribution of bottom habitats on the continental shelf off 
South Carolina and Georgia. South Carolina Department of Natural Resources. 

Van Dolah, RF, PH Wendt, and N Nicholson. 1 987. Effects of a research trawl on a hard-bottom assemblage of sponges and 
corals. Fisheries Research 5(1): 39-54. 

Vaughan, ND, TC Johnson, DL Mearns, AC Hine, WW Kirby-Smith, JF Ustach, and SR Riggs. 1987. The impact of Hur- 
ricane Diana on the North Carolina continental shelf. Marine Geology 76(1-2): 169-176. 

Wagner, LM. 2006. Population genetic structure of the temperate scleractinian coral, Oculina arbuscula, in coastal Georgia. 
Master of Science, Georgia Southern University, Statesboro, GA. 

Weins, JR. 1976. Population responses to patchy environments. Annual Review of Ecology and Systematics 7:81-120. 

Wenner, EL, DM Knott, RF Van Dolah, and J Burrell, VG. 1983. Invertebrate communities associated with hard bottom 
habitats in the South Atlantic Bight. Estuarine Coastal and Shelf Science 17:143-158. 




CHAPTER 3: CHARACTERIZATION OF THE MARINE DEBRIS 

3.1 INTRODUCTION 

The accumulation of debris in the marine environment is an increasing problem worldwide. Marine debris is aes- 
thetically displeasing, can be a nuisance to boaters and the shipping industry, and can negatively impact marine 
biota (Derraik 2002). The abundance and spatial distribution of marine debris is dependent upon several factors, 
including its origin/source (e.g., terrestrial vs. maritime), ocean currents, wind patterns, and physiographic char- 
acteristics (Galgani et al. 2000; Donohue et al. 2001 ). Depending upon their composition, individual debris items 
may persist for a long time in the marine environment. In particular, plastics, which are the dominant debris type 
in numerous marine systems (Derraik 2002), tend to break down slower in the ocean than items on land due to 
lower temperatures and fouling by marine organisms (Andrady 2000). 

Derelict fishing gear is a common debris type of maritime origin, and areas with concentrated fishing activity may 
contain elevated amounts of debris (Hess et al. 1 999; Galgani et al. 2000). Derelict fishing gear and other marine 
debris can impact environments in several ways. Floating debris may facilitate the spread of non-native species 
to new areas by providing a means of transportation (Aliani and Molcard 2003). Plastic items are often ingested 
by or entangle marine organisms, including fish, seabirds, sea turtles and marine mammals (Laist 1997). Lost 
fishing gear, such as monofilament nets and traps, may affect marine organisms both by direct injury to benthic 
habitats and organisms (Donohue et al. 2001) and by continuing to catch fish and invertebrates ("ghost fishing", 
Dayton etal. 1995). 

Although smaller in size than large monofilament nets, hook and line is a prevalent gear type, particularly among 
recreational fisheries, and can also be detrimental to marine organisms (Chiappone et al. 2005). Effort is often 
concentrated at popular fishing sites, and consequently hook and line fishing may affect small areas but also 
inflict a high amount of damage within the affected areas (Asoh et al. 2004). Fishing line entangles readily in 
coral, which may lead to progressive fouling by algae and eventually, coral death (Schleyer and Tomalin 2000; 
Yoshikawa and Asoh 2004). Chiappone et al. (2005) documented numerous cases of tissue abrasion in branch- 
ing gorgonians, milleporid hydrocorals, and sponges in the Florida Keys NMS. 



A key challenge to marine debris mitigation is effectively assessing the distribution and density of debris in order 
to prioritize removal efforts. Traditional methods such as beach surveys are economical, widespread, and easy 
to implement (Rees and Pond 1995), but are limited to intertidal areas. Other survey techniques, including in 
situ underwater surveys, trawling, aerial flyovers, and remote sensing represent innovative approaches that are 
gaining increasing utilization. For example, SCUBA diving surveys have been employed to describe the distri- 
bution of derelict fishing gear in coral reefs within the Florida Keys National Marine Sanctuary (Chiappone et 
al. 2004) and in the Northwestern Hawaiian Islands in conjunction with cleanup efforts (Donohue et al. 2001). 
Using oceanographic modeling and remote sens- 
ing techniques, the "GhostNet" program in the Pa- 
cific Ocean is developing tools to track debris and 
identify likely locations of accumulation of derelict 
fishing nets and other marine debris (http://www. 
highseasghost.net, accessed May 16, 2006). 

Understanding the sources and processes that 
drive marine debris distribution patterns is crucial 
to remediation efforts. High concentrations of de- 
bris are often found in areas of concentrated fish- 
ing activity, shipping channels, or riverine outflows 
(Galgani et al. 2000). Further, wind and currents 
can transport debris to areas far from the original 
location of loss or dispersal. For example, oceanic 
convergence zones in the North Pacific Ocean are 
thought to contribute to the accumulation of de- 
bris in the NWHI (Donohue et al. 2001; Donohue 

2005). 

Image 14. Net entangled in live bottom. 





Recently, there has been increased concern about the potential accumulation of marine debris in Gray's Reef 
National Marine Sanctuary (GRNMS). Since the establishment of GRNMS in 1 981 , the population of neighboring 
coastal counties has increased substantially (-40% from 1980-2000), and has been forecast to increase an addi- 
tional 32% by 201 5 (CGRDC2006). Coincident with this population increase, the use of the sanctuary for boating 
and fishing activities has also increased. In 1983, aerial flyovers documented 106 vessels in the sanctuary during 
62 overflights (1.7 boats/flight); in 1999, 527 boats were observed during 90 overflights (5.9 boats/flight) (NOAA 
2006). While most commercially employed gear has been prohibited in the sanctuary, Gray's Reef is a popular 
recreational fishing site both for king mackerel and bottom fish such as red snapper, grouper, amberjack, and 
especially black sea bass. Hook and line is the dominant gear type to target these species, although spearfishing 
with non-power spearheads is also conducted (NOAA 2006). Several sport fishing tournaments take place off 
of the Georgia coast each year, with Gray's Reef being a premier location (Ehler and Leeworthy 2002). Current 
regulations prohibit the deposition of most materials in the sanctuary, with the exception of fish parts, bait and 
chumming materials, effluent from marine sanitation devices, and vessel cooling water (NOAA 2006). The extent 
of external inputs from sources outside the sanctuary is unknown but also of concern. 

The characteristics of bottom features in Gray's Reef may influence the accumulation and spatial distribution of 
debris in the sanctuary. Kendall et al. (2005) estimated that the GRNMS seafloor is comprised of approximately 
75% sand, 25% sparsely colonized live bottom, and less than 1 % densely colonized rock outcroppings or ledges. 
However, despite their limited area, ledges may be most vulnerable to debris accumulation. The abundance of 
sessile benthic organisms and structurally complex features such as overhangs and caves provide ample op- 
portunities for debris items to become lodged or entangled. In addition, ledge features are targeted by fishermen 
due to the high abundance and diversity of target fishes that reside there. 

The most recent management plan calls for specific measures to assess, monitor, and remove debris from 
targeted areas within the sanctuary (NOAA 2006). Activities were proposed to a) clarify regulatory authority to 
address materials discharged or deposited outside the sanctuary, b) develop and implement a marine debris 
education and outreach program, and c) develop and implement a debris monitoring and assessment study. In 
addition, GRNMS organizes divers to remove debris from targeted locations within the sanctuary through the 
annual "Sweep the Reef, Sweep the Beach" World Ocean Day Cleanup. 

Understanding the amounts, distribution, and types of debris in the sanctuary is the first step to improving cleanup 
efforts and will aid managers in prevention and education efforts. To date, the types of debris, and its distribution 
and abundance have not been quantified. The objectives of this component of the characterization were to: 

1 . Describe the abundance, types, and spatial distribution of marine debris in GRNMS, 

2. Determine whether debris presence is associated with bottom type, 

3. Determine the causes of observed spatial patterns of debris at ledges by identifying what factors, such 
as ledge height and area and observed boat activity, are related to debris presence and abundance, 

4. Predict debris densities at unsampled locations within GRNMS, and 

5. Explore relevance of ocean currents in GRNMS to debris accumulation patterns. 

Finally, the results will be used to recommend a strategy for identifying high priority sites for targeted debris 
removal, and conducting periodic monitoring to determine rates of debris accumulation in different areas and 
bottom types within the sanctuary. 



3.2 METHODS FOR MARINE DEBRIS SURVEYS 

Sampling for marine debris occurred along a 100 m 2 transect at randomly selected sites as outlined in Chapter 1 . 
Marine debris was recorded for the entire 1 00 m 2 transect. Debris was defined as any man-made object and was 
separated into two main categories, fishing gear and non-gear. Subcategories of fishing gear were not always 
noted but included monofilament line, leaders, spear gun parts, and other/undescribed (e.g. jigs or lead weights). 
Subcategories of non-gear marine debris included cans, bottles, and other (e.g. clothing, twine, tennis ball, wood 
plank, lift bag). Rope and mesh bags were found at a few sites and were scored as non-gear even though they 
may have been associated with fishing (e.g., rope could be used to mark ledge sites and mesh bags could be 
used for chumming). Fishing line that crossed the transect, but was not completely within it, was counted as a 
single item. Monofilament line with a leader attached was counted as a single piece of gear in the leader cat- 
egory. 




3.3 METHODS FOR DATA ANALYSIS 
Quantity, types, and spatial distribution of 
debris 

Survey statistics for the quantity and types of debris 
were calculated for the entire survey domain and 
according to bottom types (Table 3.1, Table 3.2). 
Observed density of total debris, fishing gear, and 
non-gear items were entered into a Geographic In- 
formation System (GIS) and mapped according to 
geographic position of survey transect in ArcView 
v9.1. 

Effect of bottom type 

First, the hypothesis that presence of debris var- 
ies significantly by bottom type was tested. For this 
analysis, flat sand and rippled sand were combined 
into a single "sand" category due to the low num- 
ber of sampling locations in these bottom types 
compared to ledge and sparse live bottom. Bottom 

type (ledge, sand, sparse live bottom) was identified for each site based on diver observations. Debris was clas- 
sified as "present" or "absent" at each site and the presence/absence data were modeled using logistic regres- 
sion (Proc Logistic, SAS v9.1). Bottom type was included as a class variable. If the main effect was significant at 
the alpha =0.05 level, contrast statements were then constructed in Proc Logistic to test for differences in debris 
density among each pair of bottom types. 




Image 15. Miscellaneous debris items. 



Influence of ledge characteristics and boat density on debris 

Given that 90% of the observed debris was found on ledges, additional descriptive statistics were calculated and 
tests were performed to identify ledge characteristics that were associated with higher amounts of marine debris. 
The occurrence of debris on ledges according to the ledge area and height categories previously classified by 
GIS analysis (Kendall and Eschelbach 2006) were summarized in pie charts. Briefly, ledges were categorized as 
short (<58.5 cm), medium (58.5-89.2 cm), or tall (>89.2 cm) by rank ordering their heights and assigning 1/3 of 
the ledges to each category (Kendall and Eschelbach 2006). Area calculations were used to categorize ledges 
as small (<316.5 m 2 ), medium (316.5-731.4 m 2 ), or large (>731.4 m 2 ), again by assigning 1/3 of the ledges to 
each category (Kendall and Eschelbach 2006). 

Ledge characteristics that were suspected to be positively associated with debris accumulation were identified 
and included mean ledge height measured in situ, ledge area (m 2 ) based on GIS analysis, mean undercut width 
(m) measured in situ, and percent cover of benthic organisms measured in situ. Total debris (per 100 m 2 ) was 
plotted against each of these ledge variables and a non-parametric Spearman Rho rank order correlation statis- 
tic was calculated (SAS v9.1 , Proc Corr). 

An additional factor that may influence the distribution and abundance of debris is the level of fishing and boating 
activity. Positions of boats in GRNMS from 1 998 to 2004 were determined from multiple sources including national 
reconnaissance systems and entered into a GIS. Positional accuracy was within 26 m and boats were classified 
as either moving or stationary. Both stationary and moving boats from all seasons were included for two reasons. 
First, to eliminate potential bias from king mackerel fishing tournaments, the boat data was initially separated by 
season and status (stationary vs. moving) to determine whether use patterns differed among bottom and pelagic 
fishers. However, regardless of how data was partitioned, the overall spatial patterns in boat density were con- 
sistent. Second, any boat could dispose of debris at sea, regardless of whether they were bottom fishing, trolling 
for pelagics or bait, or not fishing at all. To determine how the intensity of activity varies over space, the sanctuary 
was divided into 0.25 km 2 cells (500 m x 500 m) and the number of boats within each cell was calculated. Spear- 
man correlation was used to examine the association between the number of boats and the average number of 
debris in each cell (SAS v9.1 , Proc Corr). After exploring several spatial scales (1 km 2 , 0.25 km 2 , 0.09 km 2 , and 
0.01 km 2 ), the 0.25 km 2 scale was considered most appropriate for GRNMS. The 1 km 2 scale was found to be 
too coarse to capture potentially meaningful small scale variability. Conversely, the finest scale options resulted 




in too many cells that contained no sample loca- 
tions. We used the information on boat distribution 
patterns to divide the sanctuary into areas of "low" 
versus "high" boat density. A frequency histogram 
of boats per cell was used to help determine a cut- 
off between low and high density areas. 



Next, we modeled debris data to determine if 
ledge characteristics and boat density were signifi- 
cant predictors of the presence and abundance of 
debris at ledge sites. Due to the presence of nu- 
merous sites with zero debris items, the data was 
analyzed using a two-step conditional model that 
is often used for zero-inflated data (Cunningham 
and Lindenmayer 2005). This approach separates 
variables that determine whether or not debris is 
present from variables that determine the amount 
of debris, given presence. The variables included 
the boat density (low, high), mean ledge height (m), 
ledge area (m 2 ), mean undercut width (m), and per- 
cent cover of benthic species. In the first step, the debris was treated as present or absent and the presence/ab- 
sence data were modeled using logistic regression (Proc Logistic, SAS v9.1). In the second step, only sites in 
which debris was present were considered. At sites where debris was present, the number of debris items was 
modeled with a generalized linear model (Proc Genmod, SAS v9.1) with a negative binomial distribution and a 
log link. The negative binomial variance distribution was chosen because it requires fewer assumptions than the 
normal or Poisson distribution and is appropriate for modeling skewed count distributions (White and Bennetts 
1996). A Pearson's Chi-Square test was used to assess the goodness of fit of the negative binomial model to 
the data. At both stages, only main effects were considered, and parsimonious models were selected by using 
backward elimination of non-significant variables (a=0.05). 




Image 16. Aluminum beverage container. 



Predicting debris density 

Ideally, it would be most beneficial to use the two-part conditional model to generate a map of expected debris 
density at ledges throughout GRNMS. However, estimates of the covariates undercut width and percent cover 
were not available for all ledges. Instead, we used ordinary kriging, an interpolation method, of observed debris 
data to predict debris density at ledges throughout GRNMS. The procedure is based only on the observed values 
and does not explicitly take into account other variables that were found to influence debris patterns. In this ap- 
proach, the spatial covariation among all possible sample points is used to develop an estimate of debris density 
at unsampled locations, based on appropriate weightings of observed values at neighboring sites. Variogram fit- 
ting and ordinary kriging were conducted in SAS to make predictions over a 25 m grid scale, which was deemed 
to be an appropriate scale based on the size distribution of ledges within GRNMS. The interpolated debris den- 
sity was mapped in ArcView v9.1 for cells that were completely within or crossed the boundary of a ledge. In 
addition, a 25 m buffer was added to each ledge to aid in visualization of the predicted values. 

Ocean currents at GRNMS 

Finally, the potential influence of tidal currents on debris distribution was explored. Tidal currents may not only 
affect the patterns of debris that originated within GRNMS, but may also be responsible for depositing debris at 
GRNMS that originated elsewhere. Ocean current profile data from September 1, 2005 to February 28, 2006 
were obtained from the NOAA Station 41008 buoy located in GRNMS (http://www.ndbc.noaa.gov/, accessed 
March 23, 2006). Current speed and direction was measured at hourly intervals with a surface Acoustic Dop- 
pler Current Profiler (ADCP) mounted to the buoy. The direction the current is flowing is measured from 0-360°, 
where 360° is due north, and 0° means that no current was measurable. Measurements were made at 1 m inter- 
vals from surface to bottom. The general direction of bottom currents were examined by plotting the frequency of 
direction measurements at 15 m depth, the deepest reliable measurements available for GRNMS. 




3.4 RESULTS 

Quantity, types, and spatial distribution of 

debris 

A total of 93 items were found during field sur- 
veys at GRNMS. Debris was present at 32 out of 
the 179 survey sites. The number of debris items 
found within a 100 m 2 transect ranged from to 10 
items. Approximately two-thirds of all observed de- 
bris items were fishing gear, and about half of the 
fishing-related debris was fishing line (Table 3.1). 
Other fishing related debris included leaders and 
spear gun parts. Non-gear debris included cans, 
bottles, and rope. Other debris, classified as non- 
gear, included such items as wood, electrical wire, 
and a pair of pants pockets. The spatial location 
of total observed debris and fishing gear density 
are shown in Figure 3.1a-c. Highest incidence of 
debris occurred at ledges in the center of the sanc- 
tuary. 




Image 17. Weight belt used for SCUBA diving. 



Effect of bottom type 

Out of the 32 sites where debris was present, all except three sites were classified as ledge bottom type (Table 
3.2, Figure 3.2). Only non-fishing gear items were found on sand bottom types. Results from the logistic regres- 
sion indicated that the presence of debris varied significantly by bottom type (Table 3.3). The probability of debris 
presence on ledges was 0.32 (n = 92), which was significantly greater than the probability of presence on sparse 
live bottom (0.02, n = 50) and sand bottom types (0.06, n = 37). There was no significant difference in the pres- 
ence of debris between sparse live bottom and sand. 

Influence of ledge characteristics and boat density on debris 

The highest debris counts were encountered on ledges that had previously been classified as "tall" in height 
and "large" in area (Figure 3.3). The number of debris items present was positively correlated with ledge height 
(Spearman's Rho = 0.44, Figure 3.4). Debris was present at all of the five tallest ledges surveyed. In addition, the 
number of debris items increased with increasing percent cover (Spearman's Rho = 0.48, Figure 3.5), undercut 
width (Spearman's Rho = 0.41, Figure 3.6) and ledge area (Spearman's Rho = 0.40, Figure 3.7). 

The number of boats per 0.25 km 2 cell ranged from 0-99, with higher boat densities observed in the central part 
of the sanctuary (Figure 3.8). In nearly half of the sanctuary (1 07 out of 234 cells), no boats were recorded, while 
in much of the remaining cells, only a few boats were observed. A natural break in frequency of cells occurred 
between density classes 4 and 5 (Figure 3.9a). Only 33 cells had an estimated density of >5 boats, and further, 
these cells were clustered in the center of the sanctuary. Therefore, cells with <5 boats were labeled as having 
low boat density and cells with >5 boats were labeled as having high boat density (Figure 3.9b). 



The number of boats per 0.25 km 2 cell 
was positively correlated with the ob- 
served average debris density (num- 
ber/100 m 2 ) (Spearman's Rho = 0.42, 
Figure 3.10). The majority of debris 
items (80 out of 93) were found in the 
area that was defined as having high 
boat density (>5 boats/0.25 km 2 cell), 
even though more than twice as many 
sites were sampled in the region of low 
boat density (n = 122 compared to n = 
57, respectively). In addition, the corn- 



Table 3.1. Frequency of debris types pooled across all GRNMS survey sites (n=179). 



Debris type 


Total number 


% of total debris 


Fish line 


31 


33.3 


Leader 


10 


10.8 


Spear gun parts 


1 


1.1 


Non-descript/Other gear 


21 


22.6 


Total gear pieces 


63 


67.7 


Cans 


14 


15.1 


Bottles 


2 


2.2 


Rope 


4 


4.3 


Other 


10 


10.8 


Total non-gear 


30 


32.3 


Total debris 


93 


100 







S fSv - 'T -At"'-' 
jii- I 5- I Kilometers 

0.5 1 




© ^ 

(T) '"" e •*." 

of #. ^ >,A o^» ^ n ° - o , o 

' O , a . 



o>° c , 



O o 



:■ ' O 
O 






O -o v 













o °«b 0° 



:, * 9 



° 



> 9 



°, '• 



o P 

/ o 

5 



•• "A" 



Flat sand 
Rippled sand 



Sparsely colonized live bottom 


Observed # total debris 


O 


3-4 


Densely colonized live bottom 





O 


5-6 


GRNMS Boundary 


• 1-2 


O 


7-10 



Figure 3.1a. Spatial distribution of total debris (number per 100 m 2 ). 




Flat sand 
Rippled sand 



Sparsely colonized live bottom 
Densely colonized live bottom 
GRNMS Boundary 



Observed # gear Q 3.4 

00 Q 5-6 



o 1-2 



o 



7-9 



Figure 3.1b. Spatial distribution of fishing gear (number per 100 m 2 





! I -f I Kilometers 




Flat sand 
Rippled sand 



Sparsely colonized live bottom 
Densely colonized live bottom 
GRNMS Boundary 



Observed # non-gear f -| _ 2 

° a , 



Figure 3.1c. Spatial distribution of non-gear (number per 100 m 2 



Table 3.2. Presence and average number of debris items per 100 m 2 by bottom type. 



-<►. 



Bottom type 


Total Sites 


Number of sites 
with debris 


Total # 
debris 


Average # debris 
(±SE) /100m 2 transect 


Ledge 

Sparse live bottom 
Flat sand 
Rippled sand 


92 

50 
22 
15 


29 

1 

2 


89 
2 

2 


0.97 (±0.21) 
0.04 (±0.094) 

0.00 
0.13 (±0.09) 


All 


179 


32 


93 


0.52 (±0.11) 



Table 3.3. Contrast estimates, standard errors, and chi-square statistics from logistic regression of presence of marine debris in 
GRNMS by bottom type. 



Variable 


df 


Parameter 


SE 


95% Wald Confidence 


Wald Chi-Square 


Pr>ChiSq 






Estimate 




Limits 








Lower 


Upper 


Bottom type 


2 


- 




- 


- 


15.5 


0.0004 


Ledge vs. Sparse 


1 


3.14 


1.03 


1.11 


5.16 


9.2 


0.002 


Live Bottom 
















Ledge vs. Sand 


1 


2.06 


0.76 


0.57 


3.55 


7.3 


0.007 


Sand vs. Sparse Live 


1 


1.08 


1.24 


-1.36 


3.52 


0.8 


0.386 


Bottom 

















E 

o 
o 



a. 



a> 
■a 

(D 

u> 

re 

(D 

> 
< 



i.tu 
1.20 - 








1 nn - 












0.80 - 










0.60 










0.40 - 










0.20 - 






T 


T i 


00 - 






, ' T i 


i 1 i 



Sand 



Figure 3.2. 

type. 



Ledge Sparse Live Bottom 

Average # debris items (±SE) per 100 m 2 transect by bottom 



position of debris varied between the two 
regions. For example, 75% (60 out of 80 
items) of the debris in the high boat density 
area was fishing gear. In comparison, 23% 
(3 out of 1 3 items) of the debris found in the 
low boat density area was fishing gear. Two 
out of the three fishing gear items found in 
this region were observed at a site just out- 
side the boundary of the high density boat 
area. 

Results from the two-part conditional mod- 
el indicate that multiple characteristics of 
ledge features and boat density influence 
observed distribution patterns of debris. 
Boat density was a significant predictor for 
presence of debris and abundance of de- 
bris, given presence (Table 3.4). Additionally, 

ledge area and percent cover were significant predictors of presence of debris, and mean ledge height was a 
significant predictor of abundance of debris, given presence. There was a higher probability of encountering 
debris with increasing ledge height. However, at both stages of the model, the estimates for the significant ledge 
variables were small. The Pearson Chi-Square test statistic indicated that the negative binomial distribution was 
appropriate. The null hypothesis of this test was that the data fit the model, and we were unable to reject this 
hypothesis (X 2 = 28.05, df = 26, p = 0.356). 

Predicting debris density 

Interpolated density of debris at ledges is displayed in Figure 3.11 . Highest predicted densities of up to 8.4 piec- 
es/100 m 2 occurs in the center of the sanctuary where highest numbers of debris items were observed. Moderate 
amounts of debris are expected to occur in regions east, west, and south of this area. In much of the sanctuary, 
a density of zero items/ 100 m 2 is predicted. 

Ocean currents at GRNMS 

Direction of bottom currents was bimodal (Figure 3.12a), consistent with an ebb and flow tidal cycle. Arrows 
pointing in the direction of the two modes, 305 and 125 degrees, were overlaid on the GRNMS map (Figure 
3.12b). Thus, the tidal currents appear to flow in a SE-NW direction, perpendicular to the orientation of the north- 
ernmost line of ledges. Although we did not have enough information to test for the significance of currents on 
debris distribution, this pattern indicates that currents may be favorable to accumulation of debris at some loca- 
tions where densities were highest. For example, debris originating north of the northernmost ledge line could 
travel south over sandy bottom with the tide until it encounters a ledge. The potential role of currents in mitigating 
distribution of debris will be discussed further in the following section. 



Table 3.4. Two part conditional model for ledge bottom type to test for the effects of boat density (low, high) and ledge characteristics 
(ledge area (m 2 ), mean ledge height, mean undercut width, and percent cover of benthic organisms) on presence and abundance, given 
presence, of marine debris in GRNMS. The first stage models presence-absence with logistic regression, while the second stage predicts 
density, given presence, with a generalized linear model with a negative binomial distribution. P-values less than 0.05 were considered to 
show a significant effect, and models were reduced by backward elimination to remove non-significant variables. 



Variable 



Parameter 


SE 


Estimate 




0.65 


0.29 


0.0004 


0.00021 


0.024 


0.0089 


0.82 


0.35 


0.006 


0.002 



Wald Chi-square Pr>ChiSq 



Stage 1 Boat density (high vs. low) 

Ledge area 
Percent cover 

Stage 2 Boat density (high vs. low) 

Ledge height 



5.1 
3.94 
7.79 

5.41 
6.29 



0.024 
0.047 
0.006 

0.020 
0.012 




Short 
(<58.5) 



E 
o 

*j Medium 

-g, (58.5-89.2) 

(D 

I 



Tall 
(>89.3) 



Small 
(<316.5) 



n=0 



Area (m 2 ) 

Medium 
(316.5-7314) 




n=1 




n=2 



n=6 






n=1£ 




n=1 



Large 
(>731 .4) 




n=7 



Observed # 
debris/1 00m 2 



n=7 



D5 

D6 



n=50 



Figure 3.3. Frequency of debris at ledge sites by area and height class combinations as de- 
termined by Kendall and Eschelbach (2006). See methods for further description. The number 
of ledges within each area-height combination is noted by n=i. 



12 -, 




12 -, 



50 100 150 

Mean ledge height (cm) 



200 



Figure 3.4. Relationship of observed number of debris items (per 
100 m 2 ) with ledge height (measured in situ). 




40 60 

% Cover 

Figure 3.5. Relationship of observed number of debris items (per 
100 m 2 ) with percent cover of benthic organisms (measured in 
situ). 



12 7 




50 100 150 200 250 

Mean undercut width (cm) 



300 



Figure 3.6. Relationship of observed number of debris items 
(per 100 m 2 ) with undercut width (measured in situ). 




2000 



4000 
Ledge area (m ) 



6000 



8000 



Figure 3.7. Relationship of observed number of debris items 
(per 100 m 2 ) with ledge area (determined from GIS analysis). 




3.5 DISCUSSION 

The purpose of this study was to characterize debris 
patterns in GRNMS to support cleanup and monitor- 
ing of debris in the sanctuary. To our knowledge, this 
was the first study to quantify the types and amount 
of debris in offshore waters Georgia. A variety of de- 
bris items, including plastics, Styrofoam products, 
metal, glass, and fishing-related items, have been 
observed and removed during beach surveys in 
coastal Georgia (Gilligan et al. 1992). While fishing 
gear constituted a small portion of the total debris 
found on the beaches, Gilligan et al. (1992) noted 
that the impact of small items such as fishing line 
and string may have a disproportionately large effect 
due to the potential for entanglement of the benthic 
substrate, organisms, and other debris items. In con- 
trast, in terms of number of debris items, fishing gear 
was more common than consumer related items 
(e.g., bottles, cans, packaging, etc.) in GRNMS, 
which is not surprising given the popularity of recre- 
ational fishing in the sanctuary. The types of debris 
observed in GRNMS are similar to those found in 
coral reef habitats in the Florida Keys National Ma- 
rine Sanctuary (FKNMS). Both sanctuaries have a 
large recreational fishing contingent. Lost hook and 
line gear is the dominant debris type in both sanctu- 
aries, although lost lobster traps also are common in 
FKNMS (Chiappone et al. 2004). 

The distribution and abundance of marine debris 
in GRNMS is related to the bottom type, the level 
of boating/fishing activity, and local characteristics 
of benthic features. There is a significantly greater 
probability of presence of debris at ledges compared 
to other bottom types. Several factors may contrib- 
ute to this observation. First, the abiotic features of 
ledges (e.g., crevices, changes in relief, overhangs) 
provide numerous places for fishing line and other 
debris to snag or become trapped. Ledges also tend 
to be densely colonized with corals, sponges, and 
other biota (Chapter 2), creating further opportunities 
for debris entanglement. For example, although as- 
sociation with corals was not routinely recorded, div- 
ers noted several instances where fishing line was 
found tangled in branches of oculinid coral. Second, 
due to the association of recreationally important 
fish species with ledges (Chapter 4), these bottom 
features are often targeted by fishermen. Even in ar- 
eas with many boat observations, there were almost 
no occurrences of debris at sand and sparse live 
bottom sites. This is probably due to the concentra- 
tion of fishing effort at ledges, and because the low 
complexity of sand bottom types is less conducive to 
debris entanglement and accumulation. 




GRNMS Boundary # Boats ' ce " 



1-4 Q ^\ 9-12 
Boat locations |o ^] 5-8 Q | 13-16 H | >20 

Figure 3.8. Locations of observed boats and density of boats per 0.25 
km 2 cell. 

120 



100 



>> 
o 
c 
a 








10 20 30 40 50 60 70 80 



90 



# boats per 0.25 km cell 
Figure 3.9a. Frequency histogram of the number of boats per cell 
in GRNMS. The red line represents the selected cutoff between low 
and high boat density. 




Low boat density (0-4) 
High boat density (5-99) 
I GRNMS Boundary 



Figure 3.9b. Regions of low (0-4 boats/ 0.25 km 2 cell) and high 
(5-99 boats/ 0.25 km 2 cell) boat density. 




£ 
o 



5 - 



3 - 



Of all tested variables, boat density had the strongest as- 
sociation with both presence and abundance of debris at 
ledges. Boat density is highest in the center of the sanctu- 
ary on a SW-NE axis, with the largest concentration occur- 
ring in the vicinity of the data buoy (NOAA station 41008); 
99 boats were observed in the cell that included the buoy. 
The high density of boats in this region is likely attributed 
to several factors. Recreational fishermen noted that the 
buoy is a popular location to catch bait and troll for king 
mackerel, and a nearby ledge attracts bottom fishers (Cap- 
tain Judy Helmey and William H. "Bing" Phillips, personal 
communication). Slightly further away from the buoy, boat 
activity is less dense but still high. Fewer boats were ob- 
served in the southern portion of the sanctuary, despite the 
presence of numerous ledges, which indicates less fishing 
occurs here compared to areas of high boat density. This is 
further supported by the difference in debris types between 

the two areas. Three-quarters of the debris items found in the region of high boat density were fishing gear, while 
debris items observed in the low density region were primarily non-fishing related. 



J2 

■o 
o 

O) 

ra 

a 
> 

< 



20 



40 



60 



80 



100 



# of boats per 0.25 km cell 



Figure 3.10. For each 0.25 km 2 cell, the relationship of aver- 
age number of debris items (per 100 m 2 ) with the number of 
observed boats. 



In addition to the strong link with boat density, patterns of debris occurrence were also related to physiographic 
features of ledges. The presence of debris significantly increased with increasing area and percent cover of 
ledges; and given presence, the abundance of debris was positively related to ledge height. It is not surprising 
that ledge area is a significant predictor of presence of debris because extensive ledges are more likely to be tar- 
geted by recreational fishermen who closely monitor their depth sounder (John Duren, personal communication). 
Once good fishing spots have been located, fishermen often return to those locations. Thus, high boat density in 
the center of the sanctuary, where many large ledges, including the five with the largest area, are located, may be 
indicative of preferred fishing spots. Fishing charts that include the GPS location of bottom features and fishing 
"hot spots" can be purchased in marina stores. Several of the features on such fishing charts (www.sstcharts. 
com, accessed June 4, 2006) are in close proximity to areas where both a large number of boats were observed 
and fishing debris was found. 

There are a couple of reasons that ledge height may not have been significant in the first stage of the analysis. 
First, numerous tall ledges occur in the southern portion of the sanctuary, and they may experience lower fishing 
pressure based on boat sighting data. Such ledges may not be well known or have not been "discovered" at all, 
which may partially explain why little debris was found on them (Captain Judy Helmey and William H. "Bing" Phil- 
lips, personal communication). The importance of 
ledge height is confirmed in the second stage of 
the model; among ledges with debris, taller ledges 
have greater concentrations. 

Interpolation (ordinary kriging) of observed debris 
data was used to predict debris density at ledges 
throughout GRNMS. Highest densities were pre- 
dicted in the central region where the most de- 
bris was found, whereas little to no debris would 
be expected in the southern region. Although this 
method appears to be an effective way to estimate 
spatial density patterns, additional independent 
samples would be necessary to perform cross-val- 
idation analysis. Potential future work could refine 
the predictive model to include information on boat 
density and bottom feature characteristics, since 
these were significantly related to debris presence 

Image 18. Lead weight used for fishing. 






^PV ' f'' 1 Kilometers 

'■~~ir 0.5 ' " «. 1 -^ 



•V 



;:■* 



4£ 







if.",.*" * *-~ ■">•". .- - * 







>Ti 





N 



Flat sand 
Rippled sand 



Sparsely colonized live bottom 
Densely colonized live bottom 
GRNMS Boundary 






1.1-2 












2.1-3 












3.1-4 






Figure 3.11. Interpolated debris density (per 100 m 2 ) in GRNMS from ordinary kriging. 

and abundance. As stated previously, this was not carried out here as estimates of some variables (e.g., percent 
cover) are not available for all ledges. 

The potential for outside sources of debris to enter the sanctuary was a subject of concern in the recent draft man- 
agement plan (GRNMS 2006). Although there is no way to verify the origin of debris items found within GRNMS, 
we examined ocean current data to evaluate the potential for debris from outside the sanctuary to contribute to 
debris accumulation in GRNMS. Currents in GRNMS are primarily tidal, as indicated by the bimodal frequency in 
direction of currents recorded by the NOAA data buoy. The directions of the dominant currents, 305° and 125°, 
suggest that debris originating in the north-central part of the sanctuary could roll easily over featureless sand 
areas during a tidal cycle until it encounters the line of ledges where highest debris densities were observed. 
This is also the portion of the sanctuary nearest land, further suggesting that currents could bring in land-based 
debris items from outside the sanctuary. However, the highest densities of debris were not distributed evenly 
along this line of ledges, as would have been expected were tidal currents depositing marine debris from outside 
the sanctuary. It is likely that most fishing-related debris originates from boats inside the sanctuary. All observed 
fishing gear consisted of permitted gear types that are known to be used in the sanctuary. The prevalence of gear 
that is not used locally is often an indication that it has traveled from elsewhere, as has been observed in the 
NWHI (Donohue et al. 2001). The net movement of water could not be evaluated from available data; however, 




350 ; 




315 ; 




280 | 




"ST 245 1 

<o 

<o 

o) 210 E 

a) 

■o 

~ 175 ; 

o 

'■5 140 E 




5 105 E 




70 : 




35 ; 




= 


I I I 







50 



100 



150 



200 



Frequency of observations 
Figure 3.12a. Frequency of current direction observations at 15m depth at the 
NOAA buoy located in GRNMS. Observations were taken at hourly intervals 
from 9/1/05 to 2/28/06. 



Seim and Edwards (Seim and Edwards 
in press) demonstrated that the NOAA 
buoy-mounted ADCP in GRNMS under- 
estimated current velocity at depth. 

The influence of ocean currents on debris 
accumulation in GRNMS warrants further 
study, particularly in relation to items that 
may be more easily transported. For ex- 
ample, many of the non-fishing related 
debris items found in GRNMS consisted 
of beer and soda cans/bottles. While the 
highest concentration of non-fishing de- 
bris was also located in the center of the 
sanctuary, these items were often pres- 
ent at sites with no fishing debris and in 
the area of lower boat density. Marking 
debris items and tracking their movement 
over time is one possible way to assess 







i n_4f£H Kilometers 

~-T o.5 -yv 





vtaMt- ***** »iVtt' * *■ tT^ 







Flat sand 
Rippled sand 



Sparsely colonized live bottom 
Densely colonized live bottom 
GRNMS Boundary 






1.1-2 












2.1-3 












3.1-4 





4.1-5 
5.1-6 
6.1-7 



7.1-8 
8.1-9 
Buoy 



Figure 3.12b. Tidal current direction overlaid on map of GRNMS. The arrows represent the bimodal peaks from the frequency 
distribution (125° and 305°). 360° is due north. 




the influence of currents on debris movement and 
accumulation. 

Debris density at GRNMS can be compared to 
estimated densities in the Florida Keys National 
Marine Sanctuary (FKNMS) found during a study 
using similar methodology (Chiappone et al. 
2004). Total marine debris in the high relief spur 
and groove and low relief bottom types in FKNMS 
were estimated as 1.15 (±0.14 SE) and 1.22 
(±0.20 SE) per 100 m 2 , respectively (Chiappone 
et al. 2004), which is slightly higher than the mean 
density observed on ledge bottom type (0.97) and 
twice as high as overall mean density (0.52) in 
GRNMS. Furthermore, the distribution of debris in 
FKNMS appears to be more widespread; debris 
was recorded at 92% of sites sampled in FKNMS 
(Chiappone et al. 2004). The differences between 
the two sanctuaries may be a reflection of the dis- 
parities in accessibility; GRNMS is further from shore and likely receives fewer fishing trips than FKNMS. How- 
ever, due to the differences in the bottom types that were sampled, it is difficult to directly compare our results 
to those in the Florida Keys. Chiappone et al. (2004) also compared hook and line density between regions of 
varying fishing pressure (no fishing, fished, and catch and release zones) but surprisingly found no significant 
differences between the three areas. The authors hypothesized that this may be due to noncompliance with 
regulations and/or the deposition of debris prior to enactment of regulations in protected zones in 1997. Simi- 
larly, it is unknown when debris that we observed 
in GRNMS was deposited. Periodic monitoring 
and removal of debris at designated sites would 

greatly improve our understanding of debris ac- ,;?,.;, 

cumulation rates in GRNMS. crnm! 88 '" ' ffii''*:'-* ■■■ 




Image 19. Fishing line entangled in coral. 






3.6 RECOMMENDATIONS FOR MANAGE- 
MENT AND MONITORING 

Information gleaned from the current analysis was 
used to devise a strategy for prioritizing cleanup 
efforts (Figure 3.13). Because the overwhelming 
majority of debris was located in densely colo- 
nized ledge habitat, ledges should be considered 
a higher priority for debris mitigation and removal. 
Second, due to the significant difference in pres- 
ence and abundance of debris between regions 
of high versus low boat density, ledges positioned 
within the area of more intense fishing pressure 
are more likely to have debris. Of the 436 mapped 
ledges in GRNMS, 156 are located within the re- 
gion of high boat density. The number of ledges 
receiving top priority for cleanup can be reduced 
further by accounting for ledge height and area, 
since the results of this study demonstrated that 
presence and abundance of debris are positively 
correlated with these variables. As described in 
the methods section, all of the ledges in GRNMS 
were classified as short, medium, or tall in height 
and small, medium or large in area. Within the 
region of high boat density, 34 ledges are clas- 



156 ledges 
within high 
boat density 
region 



H 




Selection Scheme #1 



34 ledges in upper quantile for height and 
size (tall/lg) 




nrm_£i 



$ \ 



Selection Scheme #2 

85 ledges in upper 2/3 quantiles for height and 
size (tall/lg, tall/med, med/lg, med/med) 




Figure 3.13. Recommended strategy to prioritize ledge sites within 
GRNMS for debris removal. 




sified as both "tall" and "large" and would be rec- 
ommended as the first sites to target. During the 
recent "Sweep the Reef, Sweep the Beach" World 
Ocean Day Cleanup 2006, divers removed numer- 
ous debris items, including fishing gear and bev- 
erage cans/bottles, from several sites within this 
category (Gail Krueger, personal communication). 

If sufficient resources are available, an additional 
54 ledges that fall into both the upper two quantiles 
for height and area (tall/medium, medium/large, 
medium/medium) are additional recommended 
sites for debris removal, bringing the total to 85 
ledges. After debris is removed, sites should be 
monitored periodically to measure rates of new 
debris accumulation. In addition, we would recom- 
mend expanding long-term monitoring efforts to in- 
clude several ledges that are located in the areas 
of lower observed boat densities to compare ac- 
cumulation rates. Periodic updates of boat sighting 
data will allow managers to detect any changes in 
recreational boating patterns in GRNMS. 




JiiiiMkMimfeffittMmy^ 



Image 20. Example of fouled fishing line (out of water). 



Marine debris may inflict both direct and indirect damage to biota in GRNMS. Although impacts on biota were not 
quantified as part of our study, in several instances fishing line was observed to be entangled with benthic organ- 
isms, particularly the branching corals in the family Oculinidae. Fishing line, wire, hooks, and leaders can cause 
tissue abrasion when they snag on reef organisms. Chiappone et al. (2005) documented significant relationships 
between the density of lost hook-and-line gear with the density of damaged sponges, gorgonians, and milleporid 
hydrocorals, and found a positive correlation between length of fishing line and densities of damaged gorgoni- 
ans. Once entangled, fishing line may become incorporated into the reef matrix if it is overgrown by individual 
organisms (Chiappone et al. 2005). In our study, fishing line was often fouled by algae. In time, progressive algal 
fouling of fishing line entangled in coral may lead to coral death (Schleyer and Tomalin 2000; Asoh et al. 2004; 
Yoshikawa and Asoh 2004). In GRNMS, taller ledges in particular may be most susceptible to damage because 
they tend to be most densely colonized with benthic organisms (this document). The impacts of hook-and-line 
fishing gear and other debris on benthic organisms in GRNMS and elsewhere need further study because nega- 
tive effects are likely to become more severe as use of the sanctuary increases. 







REFERENCES 

Aliani, S, and A Molcard. 2003. Hitch-hiking on floating marine debris: macrobenthic species in the Western Mediterranean 
Sea. Hydrobiologia 503(1-3): 59-67. 

Andrady, AL. 2000. Plastics and their impact in the marine environment. In: Carbajal S (ed) Proceedings of the International 
Marine Debris Conference on Derelict Fishing Gear and the Marine Environment, Honolulu, HI, pp 137-143. 

Asoh, K, T Yoshikawa, R Kosaki, and EA Marschall. 2004. Damage to cauliflower coral by monofilament fishing lines in 
Hawaii. Conservation Biology 18(6): 1645-1650. 

Chiappone, M, H Dienes, DW Swanson, and SL Miller. 2005. Impacts of lost fishing gear on coral reef sessile invertebrates 
in the Florida Keys National Marine Sanctuary. Biological Conservation 121(2): 221-230. 

Chiappone, M, DW Swanson, SL Miller, and H Dienes. 2004. Spatial distribution of lost fishing gear on fished and protected 
offshore reefs in the Florida Keys National Marine Sanctuary. Caribbean Journal of Science 40(3): 312-326. 

CGRDC (Coastal Georgia Regional Development Center). 2006. Georgia Coast 2030: Population Projections for the 10- 
county Coastal Region. Prepared by the Center for Quality Growth and Regional Development at the Georgia Institute of 
Technology. 

Cunningham, RB, and DB Lindenmayer. 2005. Modeling count data of rare species: some statistical issues. Ecology 86(5): 
1135-1142. 

Dayton, PK, SF Thrush, MT Agardy, and RJ Hofman. 1995. Environmental effects of marine fishing. Aquatic Conservation: 
Marine and Freshwater Ecosystems 5(3): 205-232. 

Derraik, JGB. 2002. The pollution of the marine environment by plastic debris: a review. Marine Pollution Bulletin 44(9): 
842-852. 

Donohue, MJ. 2005. Eastern Pacific Ocean source of Northwestern Hawaiian Islands marine debris supported by errant fish 
aggregating device. Marine Pollution Bulletin 50(8): 886-888. 

Donohue, MJ, RC Boland, CM Sramek, and GAAntonelis. 2001 . Derelict fishing gear in the Northwestern Hawaiian Islands: 
diving surveys and debris removal in 1999 confirm threat to coral reef ecosystems. Marine Pollution Bulletin 42(12): 1301- 
1312. 

Ehler, R, and VR Leeworthy. 2002. Gray's Reef National Marine Sanctuary (GRNMS): A Socioeconomic Review of Georgia's 
Marine Related Industries and Activities. U.S. Department of Commerce, National Oceanic and Atmospheric Administration, 
National Ocean Service, Silver Spring, MD. 

Galgani, F, JP Leaute, P Moguedet,ASouplet, Y Verin, ACarpentier, H Goraguer, D Latrouite, BAndral, YCadiou, JC Mahe, 
JC Poulard, and P Nerisson. 2000. Litter on the sea floor along European coasts. Marine Pollution Bulletin 40(6): 516-527. 

Gilligan, MR, RS Pitts, JP Richardson, and TR Kozel. 1992. Rates of accumulation of marine debris in Chatham County, 
Georgia. Marine Pollution Bulletin 24(9): 436-441. 

Hess, NA, CA Ribic, and I Vining. 1999. Benthic marine debris, with an emphasis on fishery-related items, surrounding Ko- 
diak Island, Alaska, 1994-1996. Marine Pollution Bulletin 38(10): 885-890. 

Kendall, MS, and KA Eschelbach. 2006. Boundary Options for a Research Area within Gray's Reef National Marine Sanctu- 
ary. NOAANOS NCCOS, Silver Spring. 

Kendall, MS, OP Jensen, C Alexander, D Field, G McFall, R Bohne, and ME Monaco. 2005. Benthic mapping using sonar, 
video transects, and an innovative approach to accuracy assessment: A characterization of bottom features in the Georgia 
Bight. Journal of Coastal Research 21(6): 1154-1165. 

Laist, DW. 1997. Impacts of marine debris: entanglement of marine life in marine debris including a comprehensive list of 
species with entanglement and ingestion records. In: Rogers DB (ed) Marine Debris: Sources, Impacts, and Solutions. 
Springer- Verlag, New York, pp 99-139. 

NOAA. 2006. Gray's Reef National Marine Sanctuary Final Management Plan / Final Environmental Impact Statement. 
NOAA NOS NMSP, Savannah, GA. 



Rees, G, and K Pond. 1995. Marine litter monitoring programs - a review of methods with special reference to national sur- 
veys. Marine Pollution Bulletin 30(2): 103-108. 

Schleyer, MH, and BJ Tomalin. 2000. Damage on South African coral reefs and an assessment of their sustainable diving 
capacity using a fisheries approach. Bulletin of Marine Science 67(3): 1025-1042. 

Seim, HE, and CR Edwards, in press. Comparison of buoy-mounted and bottom-moored Acoustic Doppler Current Profiler 
performance at Gray's Reef. Journal of Atmospheric and Oceanic Technology. 

White, GC, and RE Bennetts. 1996. Analysis of frequency count data using the negative binomial distribution. Ecology 
77(8): 2549-2557. 

Yoshikawa, T, and K Asoh. 2004. Entanglement of monofilament fishing lines and coral death. Biological Conservation 
117(5): 557-560. 







4.1 INTRODUCTION 

A comprehensive characterization of the density 
and size distribution of bottom fish and their as- 
sociated habitats has not yet been conducted for 
Gray's Reef National Marine Sanctuary (GRNMS). 
Not long after the sanctuary was designated in 
1981, Gilligan (1989) began compiling a list of 91 
fish species that were characteristic of the area 
and noted the general habitats of each species 
such as hardbottom, sand, or in the pelagic envi- 
ronment. Video and scuba surveys over selected 
ledges and other bottom types in recent years 
have greatly expanded this species list and have 
also estimated the relative abundance of some 
taxa (Parker et al. 1994, Reef Environmental Edu- 
cation Foundation). Other studies have had sites 
on ledges or hardbottom within GRNMS and have 
further characterized fish communities, general 
bottom associations, fish densities, and some sea- 
sonal differences among fish assemblages (Sedberry and Van Dolah 1984, Parker et al. 1994, Barkoukis 2006). 
Many additional studies have been conducted elsewhere in the South Atlantic Bight on fish communities over 
live bottom (Struhsacker 1969, Huntsman 1976, Miller and Richards 1980, Powles and Barans 1980, Grimes et 
al. 1982, Wenner 1983, Chester et al. 1984, Sedberry and Van Dolah 1984), sand bottom (Struhsacker 1969, 
Wenner et al. 1979a, 1979b), and shelf edge environments (Struhsacker 1969, Grimes et al. 1982, Barans and 
Henry 1984, Parker and Ross 1986, Gilmore and Jones 1992, Parker and Mays 1998, Sedberry et al. 2004, 
Quattrini and Ross 2006). Most of these studies were conducted at broad scales covering much of the South 
Atlantic Bight and examined differences in assemblage structure between inshore and offshore communities or 
latitudinal changes in biogeography. 




Image 21. Various species offish. 



Despite this abundance of studies in the South Atlantic Bight, very little quantitative analysis has been done on 
mid-shelf fish and their specific habitat associations (but see Parker et al. 1994). Hardbottom and limestone 
ledges are known to be key habitats for bottom fish in the region; however, the factors that make them attrac- 
tive to various components of the fish community have not been quantified. Even studies that have focused on 
individual species of bottom fish often do not quantify their fine-scale habitat preferences (Matheson et al. 1986, 
Mercer 1 989, Gilmore and Jones 1992, Harris et al. 2002, McGovern et al. 2005, Barkoukis, 2006). At best, gross 
ledge height has been categorized as small, medium, and large or sparsely, moderately, or densely colonized 
by sessile invertebrates and then related to fish assemblages (Parker et al. 1994, Riggs et al. 1996, Quattrini 
and Ross 2006). Although all of these studies have provided a wealth of biogeographic information on the South 
Atlantic Bight and an understanding of the general habitat associations of bottom fish, the more detailed struc- 
tural attributes of benthic habitat that control the variability in the fish community at locations like GRNMS have 
remained unknown. 



In contrast to the lack of detailed studies on hardbottom habitats in the South Atlantic Bight, much research has 
been focused on defining fine-scale habitat associations of fish in coral reef environments although often with 
conflicting results (Risk 1972, Luckhurst and Luckhurst 1978, Molles 1978, Roberts and Ormond 1987, Cha- 
banet et al. 1997, Ohman and Rajasuriya 1998, Friedlander and Parrish 1998, Gratwicke and Speight 2005). In 
these studies, the total abundance, species richness, diversity, and trophic structure of reef fish have variously 
been correlated with benthic characteristics such as rugosity, vertical relief, coral cover, and other variables. 
Various reef types and regions have been shown to have quite different relationships between fish community 
parameters and benthic characteristics and have demonstrated the need to identify the factors that drive com- 
munity structure of fish that are specific to each reef type and locale. 



Bottom type at GRNMS has been coarsely defined as sand, flat live bottom, and ledges. A key precursor to the 
present study was conducted by Parker et al. (1994) who examined the fish communities associated with these 
general bottom types as mapped by Hunt (1974), and further categorized live bottom as covered with sparse 
(<25%), moderate (25-50%), or dense (>50%) cover of sessile benthic invertebrates. An important next step 
is analysis of how fish community structure varies with habitat features measured on continuous rather than 
categorical scales. For example, how ledge height, area, undercut, percent cover of hard or soft substrate, and 
percent cover of sessile biota relate to resident fish communities at GRNMS has not been quantified. 

With the recent availability of much more detailed benthic maps of GRNMS (Kendall et al. 2005), a spatially 
comprehensive inventory offish and their associated bottom features is now possible. Necessary improvements 
to the existing inventory include comprehensive surveys of fish associated with all bottom types, estimation of 
the size structure of fish communities, analysis of the fine-scale benthic features that are associated with the fish 
community, and detailed analysis of the spatial distribution of key species. 

In addition to a detailed characterization of fish and their associated bottom types, the impact of other forces 
that shape fish communities remain largely unknown. For example, recreational fishing is a key user activity to 
understand and sustain at GRNMS. Fishing effort has presumably been on the rise at GRNMS in recent decades 
given increasing human populations for coastal Georgia, higher numbers of recreational anglers and days spent 
fishing (Ehler and Leeworthy 2002), and increasing boat density observed within the sanctuary (NOAA 2006). 
Recreational fishing with hook and line is the dominant approach at GRNMS although some spear fishing occurs 
as well. Both methods selectively target fish such as black sea bass (Centropristis striata), gag (Mycteroperca 
microlepis), and scamp (M. phenax) (Huntsman 1976, Mercer 1989). Trap fishing and bottom trawling are not 
allowed in the sanctuary (NOAA 2006). Direct and indirect effects on fish communities of several forms of fishing 
(e.g. commercial, recreational, artisanal) have been demonstrated in many parts of the world (Russ and Alcala 
1989, McClanahan 1994, Watson and Ormond 1994, Grigg 1994, Jennings et al. 1995, Jennings and Polunin 
1996, Jennings and Polunin 1997, Wantiez et al. 1997, Chiappone et al. 2000, Westera et al. 2003, Dulvy et al. 
2004). Despite an abundance of studies elsewhere, little is known about the effects of recreational fishing on the 
overall species richness, diversity, and abundance of benthic fish communities in the South Atlantic Bight. The 
impact of recreational fishing directly on target species has been considered at broader scales but less is known 
about impacts at discrete localities such as GRNMS. Similarly, indirect effects of such activities on fish resources 
are not well understood. The spatial distribution of fishing effort is not uniform throughout the sanctuary. Patterns 
of boat use and marine debris such as fishing line and lures indicate that the central area of the sanctuary re- 
ceives more fishing pressure than surrounding areas (Chapter 3). This offers the potential to look for differences 
in the fish community between the heavily used and less used areas. 

The objectives of this component of the study were 
to: 1) conduct comprehensive surveys of bottom 
fish associated with all bottom types in the sanctu- 
ary using a random stratified sampling approach 
and the best available bottom maps; 2) describe 
the size structure of fish communities and key 
species of interest to the recreational fishery; 3) 
identify the fine-scale benthic features that are as- 
sociated with the fish community and key species; 
4) compare fish communities and key species in 
heavily fished versus less intensively fished areas 
of the sanctuary; 5) map the spatial distribution of 
key species; and 6) offer suggestions for future 
monitoring of bottom fish at GRNMS. 

4.2 METHODS FOR FISH SURVEYS 

Visual fish surveys occurred along a 100 m 2 tran- 
sect at randomly selected sites as outlined in Chap- 
ter 1. Once at a site, the fish surveyor attached a 

tape measure to the substrate or weighted line that 

Image 22. Diver conducting fish survey. 





Image 23. Triggerfish and tomtates. 



was used to mark the site from the surface and 
began the survey. Recall that surveys were con- 
ducted in a random direction on all bottom types 
except for ledges. On dives over ledge habitat, the 
survey was conducted along the ledge face or lip if 
undercut. This allowed fish on the underside, face, 
and top of the ledge to be surveyed. As the tape 
rolled out, the diver looked forward toward the end 
of the transect and recorded all fish species to the 
lowest taxonomic level possible within the survey 
area. To maximize time spent observing fish and 
minimize the time spent writing the data, four let- 
ter codes were used that consisted of the first two 
letters of the genus name followed by the first two 
letters of the species name. In the rare case that 
two species had the same four-letter code, letters 
were added to the species name until a difference 
occurred. If the fish could only be identified to the 
family or genus level then this was all that was re- 
corded. The number of individuals per species was 

tallied in 10 cm size class increments up to 70 cm using visual estimation of fork length. If an individual fish was 
greater than 70 cm, then a visual estimate of the actual fork length was recorded. Although the benthos was 
not altered by lifting or moving rocks or other objects, the fish surveyor moved off the center line of the transect 
temporarily to identify, enumerate, or observe fish in holes and under ledges. 

Several similar looking pairs or groups of species that were observed often moved too quickly, kept a distance 
from divers, or remained far under recesses of ledges to allow consistent identification to the species level and 
were therefore identified only to the genus level. Those species were Seriola dumerili and S. rivoliana, Pareques 
umbrosus and P. acuminatus, and Decapterus maculatus and D. punctatus. 

4.3 METHODS FOR DATA ANALYSIS 

A summary of all species observed in this characterization was created in tabular format. The probability of en- 
counter, mean abundance (+/- standard error), and biomass (+/- standard error) within a 100 m 2 transect were 
provided for each species within the four bottom types surveyed. Probability of encounter is the proportion of 
surveys in a given habitat type on which a species was observed. For species that had zero values for probability 
of encounter, abundance and biomass were left blank. No standard error is given when a species was seen on 
less than three surveys although mean abundance and biomass were calculated. Mean values are rounded to 
the nearest whole number and SE is rounded to tenths. Biomass was calculated using the length-weight relation- 
ship W = al_ b , where L is length in centimeters and weight is in grams. The mid point of each size class was used 
as the value of L. For example, if a fish were in the 10-20 cm size class its length (L) for biomass estimation was 
assumed to be 1 5 cm. Values of the terms (a) and (b) were obtained from FishBase (Froese and Pauly 2005) for 
each species. For species with more than one length-weight relationship defined, values for the study nearest 
GRNMS were used. For species with no length-weight relationship published, terms for a morphologically similar 
species were used. Analysis of seasonal differences in the fish community is limited to noting presence/absence 
by sampling month (i.e., May, August, or both). 

General differences in the fish communities among bottom types were evaluated by comparing species rich- 
ness (number of species), diversity (Shannon Index), abundance, and biomass for each bottom type. Data were 
grouped by bottom type: flat sand, rippled sand, sparse live bottom, or ledge as identified by divers at each site. 
All data were log transformed to meet normality and homogeneity of variance assumptions. ANOVA was used to 
determine if multiple means comparisons were warranted followed by Tukey tests to identify which bottom types 
differed for each variable. 



Differences in the size distribution offish among bottom types were examined with size frequency histograms. 
For these, fish abundance was averaged across all species within each 10 cm size class. 



To further evaluate the differences in community structure among bottom types, the differences in the particu- 
lar fish species that were present at each site were examined using cluster analysis. Sites were hierarchically 
clustered based on presence/absence of the 78 species found in the study using Ward's Minimum Variance 
technique. Presence/absence was used to focus the analysis on simple community membership. Patterns were 
checked for stability using other clustering procedures and by clustering based on abundances of species at 
each site with extreme outliers removed. 

The fish communities at ledges were examined in greater detail through regression, cluster analyses, multiple 
means comparisons, and GIS plots. Relationships between community structure of fish and ledge characteris- 
tics were investigated with multiple regression of the 92 ledge sites. Response variables were species richness, 
abundance, and diversity of fish. Explanatory variables included mean percent cover of sessile invertebrates, 
total height, undercut height, undercut width, and total area of ledges from benthic maps (Kendall et al. 2005). 
In the event that percent cover was significant in explaining fish community variables, abiotic cover (hardbot- 
tom, sand, etc.) and biotic cover groups (coral, sponges, etc.) were examined further in a separate analysis. In 
addition, to evaluate any differences in fish community structure due to fishing pressure, fish data associated 
with ledges located in areas of high boat density were compared with low boat density areas (see Chapter 3). 
Preliminary analysis of boat count data likely to be engaged in bottom fishing and marine debris data revealed 
that the high and low boat density areas used in Chapter 3 were good surrogates for areas receiving relatively 
higher versus lower bottom fishing pressure respectively. Backwards selection in regression models was used to 
ensure that only the most influential variables were retained in the model. Analyses were performed on untrans- 
formed data except for fish abundance which was log transformed to meet statistical assumptions for multiple 
linear regression. 

The 92 ledge sites were further examined for differences in their fish communities through cluster analysis. 
These 92 sites were hierarchically clustered based on presence/absence of the 72 species found on ledges in 
the study using Ward's minimum variance technique. Patterns were checked for stability using other clustering 
procedures and clustering based on abundance with removal of extreme values (e.g. school of 40,000 Haemulon 
aurolineatum). The 92 ledge sites were also clustered based on ledge characteristics including percent cover, 
total height, undercut height, and undercut width. 



The association between fish communities and ledge characteristics was examined further by considering the 
types of ledges with which each fish community was found. This was done by comparing the results of the cluster 
analysis for sites based on their fish community with the clusters based on ledge measurements. To facilitate this 
comparison, a scatter plot was created in which the sites clustered based on the fish data were placed on the X 
axis and the sites clustered based on the ledge data were placed on the Y axis. The intersection of each site was 

then plotted in chart space. The results from earlier 
analyses indicated four clusters based on the fish 
community analysis and four based on the ledge 
measurement analysis respectively. These clus- 
ters were used to separate the chart space into 
sixteen intersecting regions. This enabled both the 
fish community, hereafter called fish clusters, and 
the corresponding physical characteristics of ledge 
sites, hereafter called ledge clusters, to be de- 
scribed relative to each other. To check the stabil- 
ity of the resulting coincidence offish clusters and 
ledge clusters, an additional clustering procedure 
was conducted wherein ledge sites were clustered 
based on both their ledge characteristics and fish 
species present in the same analysis. 

An additional summary of all species observed on 
ledges was created in tabular format. The proba- 
bility of encounter, mean abundance (+/- standard 
Image 24. Sea robin on sparsely colonized live bottom. 




error), and biomass (+/- standard error) within each 100 m 2 transect are provided for each species within the four 
ledge types identified by the cluster analysis. 

Simple plots of species richness, abundance, biomass, diversity, and ledge clusters were produced in a GIS 
using each survey location's latitude and longitude. Sites were overlaid on benthic maps of the sanctuary and 
visually assessed for spatial patterns. 

Finally, selected fish species were examined further due to their importance to the recreational fishery includ- 
ing black sea bass (Centropristis striata), gag (Mycteroperca microlepis), and scamp (M. phenax) (Huntsman 
1976). Only those surveys conducted on ledges (n=92) were used in the analyses for these species since that is 
their preferred bottom type and is where fishermen most often target them (Duren and Helmey, pers. comm.). In 
particular, Mycteroperca species are rarely seen apart from ledges. Size frequency histograms were created for 
each species with sightings designated as either in more intensively fished or less fished areas according to boat 
density (Chapter 3). Average number of fish in each size class +/- SE was calculated. The size bin containing the 
size limit of the recreational fishery (South Atlantic Fishery Management Council 2006) is noted on these plots 
and the proportion offish above and below this value was calculated for high versus low boat density areas of 
the sanctuary respectively. Where size limits fell within our size classes, the number of fish in that class was split 
proportionally above and below the value. For example, the size limit for gag is 61 cm (24 inches total length) 
and our size bin ranged from 60 to 70 cm. Therefore, 10% of the fish observed within that bin were assumed to 
be below and 90% were assumed to be above the size limit. In addition, size frequency histograms of these three 
species were plotted for each survey site and overlaid onto the benthic map of GRNMS to examine the spatial 
distribution of the various size classes. 

In addition, these three fish species of interest were examined further for relevant relationships to ledge char- 
acteristics and other variables using data from the 92 ledge sites. The abundance and mean body length of 
C. striata was examined for relationships with ledge and other relevant variables through multiple regression. 
Variables tested were total percent cover of sessile biota, total ledge height, ledge area, and location (high or 
low boat density region). In addition, all possible two-way interactions with boat density (high versus low) were 
examined. Undercut variables were not considered in this analysis since C. striata is not observed to utilize ledge 
undercuts. Abundance was log transformed to meet statistical assumptions. 

Probability of occurrence for the two Mycteroperca species respectively, as related to ledge variables, was exam- 
ined through logistic regression. Abundances were too low to enable analysis beyond simple presence/absence. 
Observations of these species made while diving indicate that these species utilize undercut ledges. Therefore, 
relationships between the presence/absence of these species and undercut height, undercut width, ledge area, 
and location (high or low boat density region) were considered along with all possible two-way interactions with 
boat density. Mean size of these species was also examined through multiple regression with these same vari- 
ables. 

The relationship between occurrences of the two grouper species and Centropristis striata relative to each other 
was also evaluated with multiple regression. For this analysis the presence/absence of both Mycteroperca spe- 
cies was combined because they often occur together. The abundance of C. striata was analyzed as a response 
variable with presence/absence of Mycteroperca species and the significant variables predicting C. striata abun- 
dance and Mycteroperca species occurrence as independent variables (ledge area, percent cover, and undercut 
height). 

4.4 RESULTS 

Visual censuses recorded 78 fish species (or species groups) from 61 genera (Appendix A). Of the 78 species, 
45 were observed in both May (68 total surveys) and August (111 total surveys) sampling periods, 8 were only 
observed in May, and 25 were only observed in August. 

On ledge habitat, 72 of the 78 species observed in the study were found with Centropristis striata (seen on 98% 
of ledge surveys), Halichoeres bivittatus (89%), Serranus subligarius (88%), and Stenotomus species (80%) 
encountered most frequently (Appendix A). The most numerically abundant species on ledges were school- 
ing juvenile fish such as Haemulon aurolineatum (mean abundance/100 m 2 transect 931 +/- 495 SE) and De- 




Image 25. Black seabass 



capterus species (195 +/- 119 SE) which were 
seen in great abundance during August surveys 
in particular. Also quite abundant at all times were 
Pareques species (55 +/- 23 SE), C. striata (28 
+/- 2 SE), and Stenotomus species (24 +/- 3 SE). 
Pelagic schooling fish such as Caranx crysos had 
by far the highest biomass (mean biomass (g)/100 
m 2 14278 +/- 9082 SE). Bottom associated fish 
with high biomass were Pareques species (6013 
+/- 3411 SE), C. straita (4111 +/- 524 SE), Archos- 
argus probatocephalus (3041 +/- 840 SE), Mycte- 
roperca phenax (3035 +/- 884 SE), and M. microl- 
epis (2586+/- 1073 SE). 

Thirty-five out of the 78 fish species were observed 

over sparse live bottom (Appendix A). The species 

most commonly encountered were C. striata (seen 

on 98% of surveys over sparse live bottom) and 

Stenotomus species (90%). These were also the 

most numerically abundant and had the highest 

biomass of bottom associated species on this habitat type although Caranx crysos, considered a pelagic fish, 

had the highest biomass (3026 +/- 1560 SE) among all species. 

Seventeen of the 78 species observed in the study were seen over flat sand habitat (Appendix A). The most com- 
monly encountered species was Xyrichtys novacula (seen on 75% of surveys over flat sand). Pelagic schooling 
fish such as Decapterus species, Caranx crysos, and Scomberomorus maculatus were the most numerically 
abundant and had the highest biomass. Eighteen fish species were seen over rippled sand, with Xyrichtys no- 
vacula again being the most frequently encountered (seen on 88% of the surveys over rippled sand) (Appendix 
A). Flat and rippled sand shared 13 species comprised mostly of pelagics and those specializing in sand habitat 
such as flatfish and razorfish. The most numerically abundant bottom species on rippled sand were X. novacula 
(9 +/- 6.8 SE) and also Stenotomus species (69 +/- 62.4 SE) due to a large number of juveniles observed at 
some sites during the May sampling period. Stenotomus species also had high biomass on rippled sand (257 g 
+/- 161.6 SE), second only to Decapterus species (628 +/- 524.8 SE), a pelagic schooling species. 

Significant differences occurred in fish species richness, diversity, abundance, and biomass among bottom types 
(Figure 4.1). Flat and rippled sand sites had lowest values for all four variables. Ledge sites had significantly 
higher species richness, abundance, and biomass than all other bottom types. Fish diversity at sparse live bot- 
tom habitat was not significantly different from that at ledge sites. The spatial distribution of species richness, 
abundance, biomass, diversity, and ledge clusters based on the fish species present at each site all showed no 
clear clumping or other non-random pattern when plotted and visually examined within the sanctuary boundar- 
ies. 



Size frequency of all fish by bottom type revealed large differences in abundance and size structure offish com- 
munities among the four bottom types (Figure 4.2). Flat and rippled sand were populated exclusively with the 
smaller size classes offish with virtually all benthic associated fish in the 0-20 cm size classes. Fish in the 20-40 
cm size classes that were observed over sand habitats were almost exclusively pelagic schooling species such 
as Caranx crysos, Decapterus species, and Scomberomorus maculatus. Also of note, rippled sand had a large 
but highly variable occurrence of fish in the smallest size class. This was due to large schools of Stenotomus 
species observed at some sites in May 2005. Sparse live bottom sites were also dominated by the smaller size 
classes of fish although occasional observations of individual longer fish, up to 70 cm FL, such as Gymnotho- 
rax saxicola were made. Ledges had much higher abundance of fish in all size classes. Of note were immense 
schools offish in the smallest size class, primarily juvenile Haemulon aurolineatum, and highest abundances of 
the very largest fish such as serranids and Lutjanus campechanus in the 60-90 cm size classes as well as Gin- 
glymostoma cirratum up to 160 cm. 



a) 

o 

ID 
Q. 
CO 



25 - 














j 


20 - 














- 


1b - 


















; 


-- 




10 - 


■ ~ 














. .. 


b - 






1 ■ - 1 








f '.".'. i 


I ■_: I 


■ ■■ ■ 


- 




i 


1 



CD 
> 

b 



40000 - 



30000 



20000 - 



10000- 






1100000- 




■ 




900000 - 








700000 - 






to 


bOOOOO - 
300000 - 




■ 


CO 

E 




m 


100000- 
-100000 - 


"* 1 


A^jp 




1 1 1 



Flat Sand 



Rippled Sand 



Sparse Live 
Bottom 



Ledge 



Figure 4.1. Species richness, Shannon Diversity (H'), abundance, and biomass offish on 
100 m 2 surveys within each bottom type. Box plots denote median and interquartile range. 
Bars denote groups that are not significantly different from each other based on Tukey 
multiple means comparisons. 



The clustering procedure on all 1 79 of the fish survey sites identified groups of sites with similar species compo- 
sition (Figure 4.3). The inflection point on the scree plot at the bottom of the clustering dendrogram indicated the 
presence of 4 well separated clusters. Two clusters included only ledge sites with 26 and 20 sites respectively. 
Sites in these two clusters typically had a similarly high number of species (16-17) (although obviously with dif- 
ferent membership). One large cluster consisted primarily of sparse live bottom sites (49 of 94) although a large 
number of low relief ledges were also included. These sites had fewer species with an average of only 8.8 per 
site. The final cluster consisted primarily of flat and rippled sand which together accounted for 32 of its 39 sites. 



An average of only 3.4 species was present at these sites. Resulting patterns were similar whether based on 
fish abundance or other clustering techniques, adding confidence that the groupings are reliable. Plots of site 
clusters overlaid on the benthic maps revealed no spatial patterns. 

The regression of ledge variables on fish community metrics indicated that only a few ledge characteristics influ- 
enced the overall fish assemblage. Species richness and abundance offish had significant positive relationships 



E 
o 
o 

"33 
o 
c 
to 

T3 

C 
13 

< 

C 
(0 



100 
90 
80 
70 
60 
50 
40 
30 
20 
10 


150 

140 
80 
70 
60 
50 
40 
30 
20 
10 


100 
90 
80 
70 
60 
50 
40 
30 
20 
10 




Flat Sand 



Rippled Sand 



Sparse Live 
Bottom 



n 



,& ^ & <P 



1200 
1100 

100 
90 
80 
70 
60 
50 
40 
30 
20 
10 






I 






f 


Ledge 



0- 10- 20- 30- 40- 50- 60- 70- 80- 90- 100- 110- 120- 130- 140- 150- 



Fork Length (cm) 

Figure 4.2. Size frequency histograms offish by bottom type. Columns denote average number offish (+/- 
SE) within each size class. Numbers within the sparse live bottom and ledge plots denote low mean occur- 
rence values which were not discernable at this scale. 



to both average percent cover and total height (Figures 4.4-4.5). These two-variable models for fish richness 
and abundance explained 65% and 70% of the variability in the data. Analysis of abiotic and biotic cover groups 
indicated a significant relationship for these two measures of the fish community occurred with only macroalgae 
and other (mostly tunicates) cover types, the two most dominant colonizers on ledges. Fish diversity (H') had a 
significant, positive, and linear relationship with ledge area and a significant negative relationship with average 
undercut height (Figure 4.6), however, these variables explained only 12% of the variability in the data. Undercut 
width was not a significant variable in predicting values for any fish community metric. Ledges under high boat 



Cluster l^ 

26 Ledges 
Avg. = 17 
species 



Cluster 2-<; 

20 Ledges 
Avg = 1 6 sp 



Cluster 3 •/ 

41 Ledges \ 
49 Sparse LB 
3 Flat Sand 
1 Rippled 
Avg. = 9 sp. 




Cluster 4 -^ 

5 Ledges ^ 
2 Sparse LB 
18 Flat Sand 
14 Rippled 
Avg. = 3.4 sp 



Figure 4.3. Dendrogram for cluster analysis of all 179 sites based on species composition. The scree plot at bottom 
indicates that there are four well separated clusters. Dark blue denotes ledge, light blue denotes sparse live bottom, 
flat sand is represented by brown, and rippled sand is yellow. 



density (presumed to be more intensively fished) 
versus low boat density areas did not have a sig- 
nificant relationship with fish community structure 
when added to the final models except for overall 
fish abundance. Even then, while the relationship 
was significant with less fished ledges having higher 
fish abundance it only explained an additional 1 .7% 
of the variability in the data (Adjusted R 2 =0.713 rel- 
ative to 0.696 for the two variable model). 

The clustering procedure on the 92 ledge sites 
based on occurrence of fish species identified 
groups of ledges with similar species composition 
(Appendix B, Figure 4.7). The inflection point on the 
scree plot below the cluster dendrogram (Figure 
4.7) indicated the presence of four clusters. Several 
clustering procedures were tested and yielded simi- 
lar patterns as did clustering based on abundance. 
Clusters 1 and 3 were typically made up of sites 
with a higher number of species (average species 
richness 17 and 16 respectively). Cluster 2 typically 
had sites with a lower number of species (average 
8.5). Cluster 4 was made up of a single site with 
nine species and had the only observations of a 
couple of rare species, Pomacanthus paru and Og- 
cocephalus nasutus. This lone site/cluster was not 
considered further. Unlike the other clusters, sites 
in Cluster 1 appeared separated from others by vir- 
tue of never having a record of Lagodon rhomboi- 
des but often having observations of other species 
such as Chaetodipterus faber, Haemulon plumieri, 
Holacanthus bermudensis, Mycteroperca microl- 
epis, and Mycteroperca phenax. Sites in Cluster 2 
were separated from others by typically having no 
occurrence of Apogon pseudomaculatus, Archos- 
argus probatocephalus, Batistes capriscus, Hae- 
mulon aurolineatum, Pareques sp., and Rypticus 
maculatus but often with occurrence of Diplectrum 
formosum. Cluster 3 often had Apogon pseudo- 
maculatus, Centropristus ocyurus, and Haemulon 
aurolineatum. 

The clustering procedure on the 92 ledge sites 
based on ledge measurements identified groups of 
ledges with similar physical characteristics (Figure 
4.8). Ledge sites in Cluster 1 were tall and heavily 
colonized but had little or no undercut. Ledges in 
Cluster 2 were tall, heavily colonized, and had deep 
undercuts. Ledges in Cluster 3 were short, sparsely 
colonized, and had little or no undercut. Ledges in 
Cluster 4 were also short and had little undercut but 
were heavily colonized. 

Species occurrence was markedly interrelated with 
ledge type (Figure 4.9). The chart of ledge sites 




■*»*&£? 



Predicted Richness = 6.77 + 0.104(Percent Cover) + 0.044(Total Height (cm)) 
R 2 = 0.66 



Parameter Estimates 

Term 

Intercept 

Mean Percent Cover 

Mean Total Height (cm) 



Estimate Std Error t Ratio Prot»|t| 

6.77 0.55 12.29 <.0001 

0.10 0.01 7.88 <.0001 

0.04 0.01 3.07 0.0028 



Figure 4.4. Multiple regression model of species richness offish 
at ledge sites. 




g/rrf 11 



Predicted (log)abundance = 3.80 + 0.022(Mean Percent Cover) 
+0.021 (Mean Total Height (cm)) 



Parameter Estimates 










Term 


Estimate 


Std Error 


t Ratio 


Prob>|t 


Intercept 


3.80 


0.14 


26.84 


<.0001 


Mean Percent Cover 


0.02 


0.003 


6.61 


<.0001 


Mean Total Height (cm) 


0.02 


0.004 


5.76 


<.0001 



Figure 4.5. Multiple regression model of (log) abundance of fish at 
ledge sites. 



■a 



■o 




I I 'S 



Wtfc. 



r») 



,..■■' 






Predicted H' = 0.64 + -0.007 (Mean Undercut Height (cm)) + 0.000(Ledge Area) 
R 2 =0.12 



Parameter Estimates 

Term 

Intercept 

Mean Undercut Height (cm)) 

Area (m z ) 



Estimate Std Error t Ratio Prot»|t| 

0.64 0.04 17.75 <.0001 

-0.007 0.002 -3.31 0.0013 

0.00004 0.00002 2.26 0.0261 



Figure 4.6. Multiple regression model of diversity offish at ledge sites. 



plotted using fish clusters versus ledge clusters re- 
vealed only 5 distinct combinations offish commu- 
nity and ledge characteristics (Areas A-E in Figure 
4.9) out of a possible total of 16 (recall there were 
four clusters of sites based on fish community and 
four based on ledge measurements). The largest of 
these, Area A, represented the intersection of sites 
in Cluster 2 based on the fish community and Clus- 
ter 3 based on the ledge characteristics. Of the 48 
sites in fish Cluster 2, all but 3 were within Ledge 
Cluster 3. Similarly, of the 51 sites in Ledge Cluster 
3, all but 6 were within fish Cluster 2. Sites in this 
area had short ledges with little or no undercut and 
consequently lacked many of the undercut associ- 
ated species such as Apogon pseudomaculatus, 
Pareques sp., and Rypticus maculatus. This area 
had high occurrence and/or biomass of Stenoto- 
mus sp. and Urophycis earlli but is perhaps better 
described by those species that were seldom pres- 
ent in contrast to other ledge clusters. In addition 
to those undercut associated species mentioned, 
Archosargus probatocephalus and Batistes capris- 
cus also had low occurrence. Area B represented 
all of the sites from Ledge Cluster 2 and 6 of the 26 
sites in Fish Cluster 1. The tall, heavily colonized, 
and deeply undercut ledges of this area were 
characterized by some of the largest fish species. 
Typically Archosargus probatocephalus, Calamus bajonado, Caranx crysos, Chaetodipterus faber, Diplodus hol- 
brookii, Haemulon aurolineatum, Haemulon plumieri, Holacanthus bermudensis, Mycteroperca microlepis, M. 
phenax, Pareques sp., and Rypticus maculatus had high occurrence and/or biomass. Area C included all but 4 of 
the remaining 20 sites in Fish Cluster 1 . The tall, heavily colonized, but less undercut ledges of Cluster 1 typically 
had high occurrence and/or biomass of Archosargus probatocephalus, Lagodon rhomboides, Balistes capriscus, 
Caranx crysos, Pareques sp., and Seriola sp.. Area D included half of the sites in fish Cluster 3 and two thirds 
of the sites in ledge Cluster 4. Sites in this area were characterized by short ledges with low or no undercut, 
low or moderate height, and high cover of sessile biota. The short but heavily colonized ledges of this area typi- 
cally had high occurrence and/or biomass of Centropristus ocyurus, Halichoeres caudalis, Microgobius carri, 
and Stenotomus sp.. Last, area E included many 
sites in Fish Cluster 3 and Ledge Cluster 1. The 
tall, heavily colonized, but less undercut ledges of 
Cluster 1 typically had high occurrence and/or bio- 
mass of Archosargus probatocephalus, Lagodon 
rhomboides, Balistes capriscus, Caranx crysos, 
Pareques sp., and Seriola sp. The stability of these 
groups was checked by clustering sites based on 
both species present and ledge variables at the 
same time. The scree plot from this indicated the 
presence of five well separated clusters that corre- 
sponded well to areas A, B, C, D and E combined, 
plus one outlier site. 

Size-frequency histograms of the key species tar- 
geted by bottom fishermen, (Centropristis striata, 
Mycteroperca microlepis, and M. phenax) revealed 
that many fewer fish were observed in size class- 
es above the size limit of the fishery (Figure 4.10). 

Image 26. School of Pareques sp. 




This was evident in both the high and low boat density portions of the sanctuary although the differences were 
markedly smaller in the low boat density areas for M. microlepis and M. phenax. The modal size for C. striatus 
was -20 cm for fish in both areas. In contrast, M. microlepis had a mode size of -40 cm in areas with high boat 
density and a higher mode size of -60-70 cm in areas with lower boat density. Similarly M. phenax had a mode 
size of -40 cm in areas with high boat density and a larger mode of -50 cm in areas with low boat density al- 
though the distribution was much flatter with fewer fish in the low boat density areas. 

The size frequency plots of C. striata for each ledge site indicated no clear clumping of this species (Figure 
4.11). In contrast, M. phenax and M. microlepis were seen at only a few sites in two main areas of the sanctuary 
(Figures 4.12-4.13). Many were observed on the tall ledges in the north/central part of the sanctuary. Another 



Cluster 1 : Average of 1 7 species per site 
Generally present: 

Calamus calamus 

Caranx ruber 

Chaetodipterus faber 

Decapterus sp. 

Epinephelus mono 

Haemulon plumieri 

Holacanthus bermudensis 

Mycteroperca microlepis 

Mycteroperca phenax 

Seriola sp. 
Generally absent: 

Archosargus rhomboidalis 

Stenotomus sp. 



Cluster 2: Average of 8.5 species per site 
Generally present: 

Diplectrum formosum 
Generally absent: 

Apogon pseudomaculatus 

Archosargus probatocephalus 

Batistes capriscus 

Haemulon aurolineatum 

Pareques sp. 

Parablennius marmoreus 

Rypticus maculatus 



r 



< 



Cluster 3: Average of 16 species per site 

Generally present: 

Apogon pseudomaculatus 
Centropristis ocyurus J 

Haemulon aurolineatum "*n 



Cluster 4: 9 species at this site 

Only occurrences of: 
Pomacanthus paru 
Ogcocephalus nasutus 



^ 



Ll 



s- 



p- 




Figure 4.7. Dendrogram for cluster analysis of the 92 ledge sites based on species composition. The scree 
plot at the bottom indicates that there are four well separated clusters. The numbers representing each site 
denote species richness. Species typically present or absent that separate each cluster from the others are 
noted at left. 



concentration was found on the tall ledges along or near the south/central boundary of the sanctuary. C. striata 
was much more abundant than the Mycteroperca species. The abundance of C. striata was significantly related 
to only percent cover and ledge area. Abundance was positively related to percent cover and negatively related 
to ledge area (Figure 4.14) although these two variables accounted for only 15% of the variability in the data. 
None of the variables tested was a significant predictor of mean C. striata size. 

The presence/absence of M. microlepis was significantly related only to the undercut height of ledges (Figure 
4.15). Presence/absence of the other grouper species, M. phenax, was significantly related to undercut height 
of ledges and ledge area although comparisons with a reduced model indicated that ledge area explained only 




High Biotic Cover (%) 74 +/- 4.9 

Moderate Total Ht. (cm) 48 +/- 2.9 
Small Undercut Width (cm) 37 +/- 7.6 
Small Undercut Ht. (cm) 16 +/- 1.4 




Cluster 4: 12 sites 



High Biotic Cover (%) 70 +/- 2.6 

Low Total Ht. (cm) 1 7 +/- 2.3 

Small Undercut Width (cm) 37 +/- 7.6 
No Undercut Ht. (cm) 2 +/- 0.8 



_Q_ 



1 



Cluster 3: 51 sites 



: 



Low Biotic Cover (%) 16+/- 2.6 

Low Total Ht. (cm) 12+/- 1.2 

No Undercut Width (cm) 4 +/- 0.9 

No Undercut Ht. (cm) 2 +/- 0.4 




r 



< 



^ 



> 



-< 



r 



p- 



^ 



^ 



i> 



5=h 



^y 




High Biotic Cover (%) 89 +/- 7.0 

Highest Total Ht. (cm) 115+/- 18.9 

Deep Undercut Width (cm) 175 +/- 38.5 
Tall Undercut Ht. (cm) 38 +/- 2.1 



Figure 4.8. At right is the dendrogram for cluster analysis of the 92 ledge sites based on ledge mea- 
surements. The scree plot at the bottom indicates that there are four well separated clusters. Descrip- 
tive and quantitative characteristics (mean +/- SE) typical of ledges within each cluster along with a 
cross-sectional cartoon are noted at left. 



8% of the variability in the data (Figure 4.15). No other variables or interactions were significantly related to 
the presence/absence of either species including heavily/less fished areas. None of the variables tested were 
significantly related to mean size of either species. The abundance of C. straita was significantly related to the 
presence/absence of the two grouper species (Figure 4.16). When Mycteroperca species are present, the abun- 
dance of C. straita is significantly lower although only 17% of the variability in abundance was explained by the 
model. 



Sites clustered based on fish community 
2 




Figure 4.9. Intersection of sites clustered based on ledge measurements and fish communities. Groups of sites 
have both the same ledge characteristics and species composition. 



4.5 DISCUSSION 

This study provides a comprehensive baseline assessment of the fish communities and their associated bottom 
types at GRNMS. Prior to this, knowledge of the fish community was based on a handful of sites within the sanc- 
tuary, a limited diversity of bottom types, and with marginal quantification of associations with fine scale benthic 
features. This baseline characterization provides the foundation for future monitoring to track the trajectory of 
fish communities in the sanctuary. 

Fish communities at GRNMS are closely linked to the benthic structure within it (Gilligan 1989, Parker et al. 
1994). Species richness, diversity, composition, abundance, and biomass offish all showed striking differences 



14 
12 
10 




Centropristis striata 
Fishery size limit = 25 cm 

BH: 70% 
AH: 30% 

BL: 68% 
AL: 32% 







O 

c 

03 
T3 

C 

-Q 
03 

03 
(D 



0.5 
0.4 
0.3 
0.2 

0.1 




Mycteroperca microlepis 
Fishery size limit = 61 cm 



BH: 87% 




AH: 13% 

BL: 65% 
AL: 35% 



1.6 
1.4 
1.2 

1 
0.8 
0.6 
0.4 
0.2 






W^ 



"~l 



Mycteroperca phenax 
Fishery size limit = 51 cm 

BH: 76% 
AH: 24% 



BL: 56% 
AL: 44% 



10- 20- 



30- 



40- 



50- 



60- 



70- 



80- 90- 



Fork length (cm) 



Figure 4.10. Size frequency histograms for selected bottom fish targeted by the recreational fishery 
for regions of low (blue) and high (red) boat density. Only ledge sites are included in the figure. Col- 
umns denote mean number of fish (per 1 00m 2 ) within each size class. Error bars represent standard 
error. The recreational fishery size limit for each species is noted and the shaded bars represent the 
size class in which the size limit occurs. The proportion offish above and below the size limit of the 
fishery within the areas of high and low boat density respectively are noted; BH=fish below size limit 
within the area of high boat density, AH= fish above size limit within the area of high boat density, 
BL= fish below size limit within the area of low boat density, and AL= fish above the size limit within 
the area of low boat density. Note the differences in abundance scale. 




Figure 4.11. Size-frequency plots of Centropristis striata observed at ledge sites. The tallest bar in the legend represents 39 
individuals. 

depending on bottom type. As expected, flat and rippled sand had the lowest values for these aspects of the 
fish community relative to sparse live bottom and ledge habitats. There were no differences in species richness, 
diversity, biomass, abundance, or species composition between fish over flat and rippled sand in any analysis. 
In contrast, a recent study of benthic infauna at GRNMS did not find such equality between sand types. Rippled 
sand had significantly higher diversity, species richness, and density of infauna in grab samples (Hyland et al. 
2006). It was hypothesized that differing hydrologic and disturbance regimes between these bottom types could 
result in differing infaunal assemblages. Why this would not translate to fish communities as well is not clear. 
Perhaps the greater mobility of fish relative to infauna results in a more even distribution between sand types. 
The largest difference between sand types was observed in May sampling when large schools of juvenile Ste- 
notomus species were observed at some rippled sand sites. Ledges, which have higher structural complexity, 
showed the highest values of nearly all metrics (except diversity which was the same for both ledges and sparse 
live bottom). 



Values of some fish community variables were comparable among studies. For example, Parker et al. (1994) 
recently conducted video surveys of fish within GRNMS. Their sampling design was similar to the present study 
in that survey sites were randomly placed and stratified by bottom type including ledge, three categories of live 



fe ■*.. jg* 



^HTL _ 







^ — J^fe-*- 5 . 3* 



?*pE Lg_LJ Kilometers 

HT 0.5 * " -=. 1 -:'" 



•V* 



;:■* 



I* 




_, 









Ei_ 



m- 



-x 



,^j> E 



Flat sand 
Rippled sand 



Sparsely colonized live bottom " " 

Densely colonized live bottom __^M6 

GRNMS Boundary 




Figure 4.12. Size-frequency plots of Mycteroperca phenax observed at ledge sites. The tallest bar in the legend represents 6 
individuals. 

bottom based on colonization density, and sand (no distinction between rippled and flat). Site selection was 
guided by the best bottom map of GRNMS available at the time (Hunt 1974). Surveys were conducted in August 
1 985, November 1 985, May 1 986, and August 1 986. Apart from the November survey, this seasonal distribution 
of samples allowed excellent comparison to the present study which was conducted in August 2004, May 2005, 
and August 2005. Species richness, density, and community structure of fish were compared between the stud- 
ies with November 1985 data excluded where possible. 

Overall fish density on ledges observed by Parker et al. (1994) was 8-20 fish/m 2 based on video surveys in Au- 
gust 1985 and 1986. The same month in the present study fell at the high end of this range with an average of 
21 fish/m 2 on ledges. This average included observations of very large schools offish such as Haemulon aurolin- 
eatum juveniles and many pelagics. Similarly, 55 species were observed by Parker et al. (1994) and the present 
study identified 59. Despite a similar density and overall number of species on ledges that were seen by the two 
studies, the lists of particular species seen were quite different. Over 1/3 of the species identified in each survey 
technique were not seen by the other (these comparisons generously assumed probable matches for fish identi- 
fied to species level in one study but only genus or family in the other). Specifically, visual surveys included 22 
species not encountered in the video approach, and conversely, the video surveys included 18 species not en- 
countered in visual surveys. These discrepancies included not only rarely seen species, which is not surprising, 




Sparsely colonized live bottom 
Densely colonized live bottom 
GRNMS Boundary 



Figure 4.13. Size-frequency plots of Mycteroperca microlepis observed at ledge sites. The tallest bar in the legend represents 
3 individuals. 



but also quite common ones. For example, two species of Acanthurids were routinely recorded on the video sur- 
veys but were never seen on visual transects. Similarly, on visual surveys, two very different species, the bottom 
dwelling Urophysis earlii and the pelagic Caranx crysos, were among the most common and abundant species 
encountered respectively. Differences in the bias of video versus visual survey techniques alone cannot account 
for these striking differences. The additional video survey in November may account for some of the 1 8 species 
seen in that study that were not encountered in the August and May visual surveys of the present assessment, 
however, this does not explain the 22 species seen with visual assessment but not video. Also of note, on flat 
bottom, 54 species were encountered in the video surveys but only 28 were found using visual assessments. 
This may be due to the more detailed stratification over flat live bottom that was used by Parker et al. (1994) in 
the video assessments. They surveyed sparse, moderate, and densely colonized (flat) live bottom whereas in 
the present study only sparsely colonized (flat) live bottom was sampled. It is also probable that the older, more 
general maps used to guide initial site selection by Parker et al. (1994) led to the sampling of somewhat different 
habitats and fish communities on ledges than the much more detailed maps used in the present study in which 
specific ledges were randomly chosen from the entire group within the sanctuary (Kendall et al. 2005). Despite 
these influences, the differences in species composition between these studies are striking. Rather than ex- 
plained by the differences in survey technique, seasonal effort, or sampling design, some considerable change 
in community structure between the two studies appears present. Unlike a trend toward more tropical species 



Leverage Plots 










Percent Cover 


Ledge Area 




5 -i 






5^ 






en H 

ID 

a: - 

a) « 

oi3- 

ro 

<S ■ 
> 

52- 


i a ' ■ I , "'"'l^-^' — ' " 




03 

o)3 — 

(0 



a>2- 


— — j^..'^-.'-"" ■' ' ' ■ 




"O 

c 

-° 1 - 


.- 




C 
i 


■ ■ 


"""'••--•., 


I 

o 
— n 






log. 


■ 




u i i i i i i i i i i i i 
10 20 30 40 50 60 70 80 90100110 




1 1 1 1 1 1 

1000 3000 5000 


I 
7000 


% Cover 


Ledge Area 




Summary of Fit 






RSquare 0.15 






RSquareAdj 0.13 








Root Mean Square Error 0.85 









Analysis of Variance Whole Model 



Source 


DF 


Sum of Squares Mean Square 


F Ratio 


Prob > F 


Model 


2 


10.801182 


5.40059 


7.5328 


0.0010 


Error 


87 


62.373963 


0.71694 






C. Total 


89 


73.175145 








Parameter 


Estimates 








Term 




Estimate Std Error 


t Ratio 


Prob>|t| 




Intercept 




2.881797 0.165383 


17.42 


<0001 




Percent Cover 


0.0100302 0.002774 


3.62 


0.0005 




Ledge Area 




-0.000151 0.000063 


-2.40 


0.0184 





Figure 4.14. Regression model of Centropristis striata abundance at ledge sites. 

found on deeper live bottom off North Carolina (Parker and Dixon, 1998), neither study had differing proportions 
of tropical versus temperate species nor pelagic versus benthic species, the assemblages were simply different. 
This could be due to some long term gradual changes between the 1985-1986 study and the present one which 
was based on 2004-2005 data. It could also be as a result of more random variation in recruitment success prior 
to the two studies, which may have resulted in different community composition at GRNMS on the scale of de- 
cades. Without quantitative observations during the interval between these studies and additional monitoring in 
the future, the variability and stability offish community patterns at GRNMS cannot be known. 

Another study that provided data for possible comparisons with the present study was conducted by Sedberry 
and Van Dolah (1984) using trawl surveys. They evaluated the influence of shelf position and season on fish as- 
semblages in the South Atlantic Bight in 1980. One site fell within GRNMS and three others were within the same 
shelf zone. During summer they found 48 species over live bottom, a value similar to the present findings and 
those of Parker et al. (1994). In contrast, to these studies however, the trawls in GRNMS revealed much lower 
fish density of 0.1132/m 2 . Differences in these values and those of other studies must be cautiously interpreted 
due to vastly different biases in trawl, trap, and visual assessment techniques. 



In addition to evaluating species composition by bottom type as others had done previously, the present study 
quantified the size structure of fish assemblages, a metric not recorded by other studies at GRNMS. Not sur- 
prisingly, ledges harbored the greatest abundance of fish in all size classes. Like structurally complex habitats 



Mycteroperca microlepis 



Whole Model Test 

Model 
Difference 
RSquare (U) 



LogLikelihood 
14.425299 
0.32 



DF ChiSquare Prob>ChiSq 
1 28.8506 <.0001 



Effect Wald Tests 

Source 

Mean Undercut Height (cm) 



1 " 



Nparm 
1 



DF Wald ChiSquare Prob>ChiSq 
1 17.7609622 0.0000 



0.5 



21.2 "* 
Mean Undercut Height (cm) 



Mycteroperca phenax 



Whole Model Test 



Model 


-LogLikelihood 


Difference 


30.829540 


RSquare (U) 


0.61 


Effect Wald Tests 


Source 




Ledge Area 




Mean Undercut Height (cm) 



Mean Undercut Height (cm) 
DF ChiSquare Prob>ChiSq 
2 61.65908 <.0001 



Nparm DF Wald ChiSquare Prob»ChiSq 
1 1 5.97305373 0.0145 

1 1 14.0116536 0.0002 



Q. 



0.5 



2940 



o 

CD 



13.4 



■st- 



Ledge Area 



Mean Undercut Height (cm) 



Figure 4.15. Logistic regression model of presence(1)/absence(0) of grouper species at 
ledge sites. Values at 50% probability of occurrence are highlighted. 



elsewhere, ledges offer by far the greatest diversity of niche space to support a variety of fish sizes and spe- 
cies. Sparse live bottom and both sand types offer virtually no change in substrate relief which could be used as 
structural refugia. Sparse live bottom offers only modest additional protection with its low density of gorgonians, 
sponges, and other sessile biota. 

The species composition of fish communities was quite distinct for ledges, sparse live bottom, and sand sites 
as indicated by cluster analysis. This finding was similar to that of Parker et al. (1994) who also used cluster 
analysis of the overall fish communities in the area based on video data. A possible exception to this separa- 
tion of fish community by bottom type was in Cluster 3 which was composed of a majority of sparse live bottom 
sites but also had a large number of ledge sites. Closer inspection indicated that these ledges were quite short 



(average height = 5 cm) with little or no undercut 
and relatively sparse colonization of sessile inver- 
tebrates. In fact, many of these sites when viewed 
underwater appeared more as sparse live bottom 
abutting sandy areas than true ledges. 

Many studies outside of the South Atlantic Bight 
have examined the relationship between structure 
of fish communities and benthic variables although 
with conflicting results. The relationships between 
fish abundance, species richness, diversity, and 
benthic characteristics appear to be highly local- 
ized phenomena. Different reef types among re- 
gions have been shown to have unique correlations 
between fish community parameters and benthic 
characteristics with few rules common to all sys- 
tems (Roberts and Ormond 1987, Chabanet et al. 
1997, Ohman and Rajasuriya 1998). Working in a 
variety of reef types and regions worldwide, many 
have found species richness offish to be positive- 
ly correlated with rugosity or vertical relief of the 
substrate (Luckhurst and Luckhurst 1978, Molles 
1978, Ohman and Rajasuriya 1998, Gratwicke and 
Speight 2005) although not in all systems (Roberts 
and Ormond 1987, Ohman and Rajasuriya 1998). 
At GRNMS, species richness was positively cor- 
related with ledge height and explained 66% of 
the variability in the data. This highlights the im- 
portance of vertical relief in adding niche space at 
GRNMS. The ledges are essentially the only hard 
vertical structure in the otherwise flat landscape of 
GRNMS. In other studies, species richness offish 
has also been found related to diversity of benthic 
cover (Roberts and Ormond 1987, Gratwicke and 
Speight 2005) or particular bottom types such as 
hard bottom and live coral (Parker et al. 1994, Oh- 
man and Rajasuriya 1998, Gratwicke and Speight 
2005) although, again results were inconsistent 
among regions (Luckhurst and Luckhurst 1978, 
Roberts and Ormond 1987). At GRNMS, species 
richness was also positively correlated with per- 
cent cover of sessile biota, namely macroalgae 
and 'other', the two dominant cover types. The 
three dimensional planar plot of this relationship 
indicates that either high percent cover or total 
height (or both) of ledges can be related to high 
richness values. Possible mechanisms are higher 
food resources for fish afforded by greater cover 
and enhanced structural refuge options and niche 
space offered by taller ledges. 

Fish abundance at GRNMS was also significantly 
correlated with percent cover and ledge height, 
which together explained 70% of the variability 
in the data. Links between fish abundance and 



Regression Plot 




"i — i — i — i — i — i — i — i — i — r 

10 20 30 40 50 60 70 80 90 100 
Percent Cover 



Summary of Fit 

RSquare 0.17 

RSquareAdj 0.15 

Root Mean Square Error 0.84 

Analysis of Variance Whole Model 

Source DF Sum of Squares Mean Square 

Model 2 12.206558 6.10328 

Error 87 60.968587 0.70079 

C. Total 89 73.175145 



F Ratio 

8.7092 

Prob > F 

0.0004 



Parameter Estimates 



Term 
Intercept 
p/a Myct.sp, 
Percent Cover 



Estimate 
2.3311367 
0.3320177 
0.0133319 



Std Error 
0.190883 
0.118075 
0.003218 



Least Squares Means Table 

Level Least Sq Mean Std Error 

absent 3.2326824 0.11316164 

present 2.5686469 0.18738277 



t Ratio Prob>|t| 

12.21 <0001 

2.81 0.0061 

4.14 <0001 



Mean 
3.06279 
2.96506 



Figure 4.16. Logistic regression model of presence(1 )/absence(0) of 
grouper species at ledge sites. Values at 50% probability of occur- 
rence are highlighted. 




Image 27. Sharksucker. 




Image 28. School of spadefish. 



benthic characteristics have been more difficult to 
identify in other studies with only weak or no cor- 
relation found with rugosity (Risk 1972, Luckhurst 
and Luckhurst 1978, Ohman and Rajasuriya 1998, 
Gratwicke and Speight 2005), cover of particular 
bottom types (Roberts and Ormond 1987, Ohman 
and Rajasuriya 1998), or other benthic variables 
(Risk 1972, Luckhurst and Luckhurst 1978, Rob- 
erts and Ormond 1987). Similar to the findings for 
species richness of fish at GRNMS, high abun- 
dance can be related to either high percent cover 
or total ledge height, both are not necessary. 

Diversity (H') of fish has also been elusive to link 
with benthic characteristics. Fish diversity has 
been positively correlated with benthic variables 
such as rugosity or reef height (Risk 1972, Molles 
1978, Ohman and Rajasuriya 1998) and live coral 
cover (Ohman and Rajasuriya 1998) but not with 
substrate diversity (Risk 1 972) and not in all studies 

or reef types investigated (Luckhurst and Luckhurst 1978, Ohman and Rajasuriya 1998). Unlike fish abundance 
and species richness, fish diversity at GRNMS was not significantly related to ledge height and percent cover. 
Instead, ledge area had a positive relationship and undercut height had, counter to expectations, a negative re- 
lationship with fish diversity. A large undercut would presumably allow greater niche space, number of species, 
and equitable distribution of community membership and therefore higher diversity but this was not the case. 
More detailed evaluation of the data indicated that larger undercuts coincided with the presence of large schools 
of a few species such as Pareques sp., which are regularly observed utilizing undercut ledges, and Haemulon 
sp. Such large monotypic schools actually lowered the overall values offish diversity at undercut ledges even in 
the presence of a larger number of species. 

Many studies have found fishing intensity to affect species richness, diversity, and abundance of fish commu- 
nities in many parts of the world (Russ and Alcala 1989, McClanahan 1994, Watson and Ormond 1994, Grigg 
1994, Jennings etal. 1995, Jennings and Polunin 1996, Wantiez et al. 1997). At GRNMS, more intensively fished 
versus less fished (based on high versus low boat density) ledges did not have a significant relationship with fish 
community structure when added to the final models except for overall fish abundance. Even then, while the rela- 
tionship was significant with less fished ledges having higher fish abundance it only explained an additional 1 .7% 
of the variability in the data (Adjusted R 2 =0.71 3 relative to 0.696 for the two variable model). While significant in a 
predictable direction, the influence is quite small and does not appear to be an important variable structuring fish 
abundance in this area. Based on the present findings, the relationships between fish abundance, species rich- 
ness, diversity, and benthic characteristics can be predicted from just a few easily quantifiable variables: ledge 
height, undercut height, percent cover, and ledge area. 

Characterizations of the particular fish species associated with ledges revealed five distinct combinations of 
ledge characteristics and fish assemblages. The four ledge types and four fish community types identified in the 
cluster analysis could have resulted in sixteen unique combinations of ledge characteristics and fish community. 
However, only five accounted for nearly 90% of the survey sites over ledges indicating a strong relationship 
between fish community membership and ledge type. Merely knowing the basic characteristics of a ledge such 
as total height, undercut width, and percent cover would allow good prediction of not only species richness and 
abundance offish, but also which particular fish species are likely to occur there. 



4.6 TARGETED SPECIES 

This study provides a comprehensive assessment for species of interest to recreational fishermen such as C. 
striata, M. microlepis, and M. phenax at GRNMS. Densities of these species were 0.52, 0.04, and 0.02/m 2 re- 
spectively on ledges as reported by Parker et al. (1 994). Densities found on ledges in the present study were half 
as high at 0.28 for C. striata and 0.01 and 0.02/m 2 respectively for the Mycteroperca species. Again, differences 



in these estimates may be due to several factors including the respective biases of the sampling methods, the 
inclusion of November sampling by Parker et al. (1 994), and the different base maps upon which sampling strate- 
gies were designed. However, if both of these assessments are considered to have adequately quantified fish at 
GRNMS, some real differences appear likely. 

The presence of both Mycteroperca species was most related to undercut height of ledges rather than any other 
variables such as total ledge height or biotic cover. Indeed. More M. phenax were observed in the fished area of 
the sanctuary which had better habitat for this species than the less fished area (Chapter 2). Fishing pressure 
also did not have a significant relationship with the simple presence/absence of these two species, however, it 
did appear to influence their size distributions. For example, modal size of M. microlepis was skewed by -25 cm 
toward smaller individuals in the heavily fished area. Similarly, the mode size of M. pfrenaxwas~15cm smaller in 
the heavily fished area although the size distribution was flattened in the less fished area. This pattern emerged 
despite the apparent presence of better habitat in the form of more deeply undercut ledges at survey sites in the 
fished area (Chapter 2). In fact, many fewer M. microlepis, and M. phenax were observed in size classes above 
the size limit of the fishery in all areas of the sanctuary. This could be the result of selective removal of largest fish 
due to fishing, as has been observed in other areas (Chiappone et al. 2000, Westera et al. 2003), as well as on- 
togenetic migration out of the area by large fish (McGovern et al. 2005). Also of note, the proportion of fish larger 
than the size limit was higher in low boat density areas than in high boat density areas for both Mycteroperca 
species. This suggests that, despite better habitat in fished areas, fish size in heavily fished areas of GRNMS 
appear to be lower than in unfished areas. 

The spatial distribution of both Mycteroperca species was quite clumped on ledges in the north central and south 
central regions of the sanctuary. Of the 92 ledges surveyed, only 20 had occurrences of these species with the 
majority only occurring on 10 ledges. Both species were often observed together at the same ledge and were 
rarely observed as lone individuals. 

In contrast, Centropristis striata occurred at 98% of the ledges surveyed and appeared evenly distributed through- 
out the sanctuary. Abundance was best explained by percent cover of sessile biota rather than ledge height or 
undercut variables. Indeed this species was never observed utilizing the undercut of a ledge. Interestingly, lower 
abundance of C. striata occurred when either of the large grouper (Mycteroperca) species were present. Lower 
abundance of C. striata at such sites could be due to predation by the large grouper (Matheson et al. 1986), 
avoidance of sites with large grouper, or some other mechanism correlated with these two variables. As with the 
grouper species, many fewer C. striata were observed in size classes above the size limit of the fishery. Emigra- 
tion is not thought to reduce the abundance of C. striata which are thought to stay in the same specific area for 
much of their adult life (Mercer 1989, Parker 1990, Barkoukis 2006,). 

There are several important caveats to consider in the present characterization. Surveys were conducted during 
the day. Some ledge associated species such as those in the family Haemulidae are known to undergo migra- 
tions away from ledges into surrounding sand habitats each night to feed. This will have the effect of inflating the 
biomass and species richness of sand areas each night to levels higher than those observed in the daytime sur- 
veys. Also, only visual surveys were used in this assessment and some fish species avoid divers. For example, 
Lutjanus campechanus and Lagodon rhomboides are often only seen at the limit of diver visibility, negatively 
biasing their counts, and other species probably move away prior to diver detection at all. Other sampling gear or 
survey techniques will evaluate such species better/differently (but not without their own set of biases). Only bot- 
tom fish or those pelagics that approached the bottom were surveyed. Among the most abundant species over 
all bottom types were pelagic species. However, the transect survey technique is not designed to sample pelagic 
fish effectively. Alternative techniques should be used to sample these species such as sonar, nets, and hook 
and line. Finally, the seasonal changes known to occur in the fish assemblage of this area (Sedberry and Van 
Dolah 1984, Parker et al. 1994) are not addressed in detail by this study. The higher number of species unique 
to August (25 on 111 surveys) relative to May (8 on 68 surveys) is not totally accounted for in the proportionally 
greater number of August surveys. Changes in proportional abundance are known to occur seasonally as well 
(Sedberry and Van Dolah 1984) but were not investigated here and should be evaluated with seasonally strati- 
fied sampling effort. 



4.7 RECOMMENDATIONS FOR MANAGEMENT 
AND MONITORING 

A long term strategy for quantitatively monitoring 
fish at GRNMS should be devised based on the 
present findings. A stratified-random sampling de- 
sign can maximize inference to the entire sanctu- 
ary and optimize effort to allow key comparisons 
among regions within it. At a minimum, sampling 
should be focused on randomly selected ledges 
stratified by the four ledge types identified here, 
as well as in the heavily fished versus less fished 
areas of the sanctuary. With limited monitoring re- 
sources the sampling should be conducted annu- 
ally within the same season. Given greater moni- 
toring resources, additional sampling could be 
undertaken to quantify seasonal effects and other 
strata. Additional strata of interest may include the 
four major bottom types used here, or perhaps 
comparison areas outside the sanctuary to place 
the sanctuary in a regional context. 




Image 29. Frogfish. 



The same assessment technique must be used each sampling period to simplify analysis and reliably detect 
changes in community structure in response to fishing pressure or other influences such as range changes due 
to global warming (Parker and Dixon 1998). For assessment and monitoring of bottom fish at GRNMS, visual 
transects should be used. No other survey technique provides as effective an approach given the visibility, bot- 
tom features, data needs, and logistical constraints. The most robust approach to monitoring of bottom fishes 
requires quantitative data. Species, size, and number of fish per unit area are needed for monitoring. Roving 
diver and trap surveys can provide relative per unit area measures at best. Trawl surveys, while spatially quan- 
titative, are ill suited to sampling ledges, the most important bottom type at GRNMS. Trawling can be done from 
the high to low side of a ledge, but is harmful to the encrusting benthic organisms, and fails to sample the sub- 
stantial component of the fish community that utilizes the undercut of many ledges. Visual point surveys require 
7.5 m visibility (Bohnsack and Bannerot 1 986), conditions that rarely occur at GRNMS. Point surveys also do not 
survey ledges efficiently, nor do findings extrapolate appropriately (see Parker et al. 1994). The ledge is essen- 
tially a linear feature that is best evaluated with a linear survey technique. In contrast, visual transects conducted 
along the axis of ledges meet the data requirements and logistical constraints imposed by the benthic features 
at GRNMS. Spatially quantitative, requiring only 2 m visibility, and simultaneous survey offish above and below 
undercut ledges make transects ideally suited to assessing bottom fish communities in this area. Although not 
without their own biases as mentioned earlier, transects offer the best approach for assessment and monitoring 
of fish at GRNMS. While the rationale provided here suggests that visual transects should play the dominant 
role in quantitative monitoring of bottom fish, other techniques should be used to accomplish other objectives. 
For example, pelagic fish should be evaluated with alternative approaches such sonar, nets, or hook and line 
sampling. In addition, long term datasets such as Marine Resources Monitoring, Assessment, and Prediction 
Program (MARMAP) trap sampling must continue. Despite difficulties of comparing these data to other studies, 
such long term, consistently collected datasets continued in the future will provide comparative information on 
increases and decreases in fish community variables relative to their data collected previously. 



The densities and size structure of selected fish species can be monitored through time with effort optimized to 
test particular hypotheses of interest. For example, power analysis based on the variability of the density data 
for a species of interest (or other variable) can be performed. Specifically, the sample size needed to detect a 
particular change in the fish community can be calculated and field work prioritized to meet that goal. Many varia- 
tions on monitoring and sampling design are possible. A reference describing monitoring options was recently 
completed for reef fish and provides a good place to begin such considerations (Menza et al. 2006). 



Another assessment option to consider in addition to stratified random sampling is a more complete comprehen- 
sive survey of all ledges. There are 436 ledges in GRNMS of various height and dimension. It is possible to visit 



every one over the course of a year or in a couple of field seasons. Ledges could all reasonably be surveyed 
without replacement (in the statistical sense) to obtain an understanding of the entire population of ledges. This 
would not necessarily need to include all of the variables and approaches here, but could instead focus on a 
subset of fish species or ledge variables such as those deemed significant in the present study. For example, 
evaluating a ledge's percent cover, total height, undercut width, and undercut height would establish its charac- 
teristics relative to the four ledge types identified in this study (Figure 4.8). Based on those characteristics it is 
possible to infer the species richness (Figure 4.4), fish abundance (Figure 4.5), diversity (Figure 4.6), and even 
species composition (Figure 4.7 and 4.9) of every ledge in the sanctuary. Based on this it would even be possible 
to estimate population sizes of ledge associated species. 



Additional activities should also be initiated to quantify fishing effort in different parts of the sanctuary. At present 
only relative levels of fishing effort can be inferred from the boat count data. The central area has higher fishing 
effort than the rest of the sanctuary, however, exactly what that level of effort and impact to the resource or CPUE 
may be is not quantifiable given present monitoring activities. 



REFERENCES 

Barans, CA, VJ Henry Jr. 1984. A description of the shelf edge groundfish habitat along the southeastern United States. 
Northeast Gulf Science 7:77-96. 

Barkoukis, AM. 2006. A temporal and spatial analysis offish trap catches within Gray's Reef National Marine Sanctuary, 
1993-2005. Masters Thesis. College of Charleston. Charleston, South Carolina. October 2006. 

Bohnsack, J A, SP Bannerot. 1986. A stationary visual census technique for quantitatively assessing community structure 
of coral reef fishes. Dep. Commer., NOAATech. Rep. NMFS 41,15 p. 

Chabanet, P, H Ralambondrainy, MAmanieu, G Faure, RGalzin. 1997. Relationships between coral reef substrata and fish. 
Coral Reefs 16:93-102. 

Chester, AJ, GR Huntsman, PA Tester, CS Manooch III. 1984. South Atlantic Bight reef fish communities as represented in 
hook-and-line catches. Bulletin of Marine Science 34:267-279. 

Chiappone, M, R Sluka, KS Sealey. 2000. Groupers (Pisces: Serranidae) in fished and protected areas of the Florida Keys, 
Bahamas and northern Caribbean. Marine Ecology Progress Series 198:261-272. 

Dulvy, NK, NVC Polunin, AC Mill, NAJ Graham. 2004. Size structural change in lightly exploited coral reef fish communities: 
evidence for weak indirect effects. Canadian Journal of Fisheries and Aquatic Sciences 61 :466-475. 

Ehler, R, VR Leeworthy. 2002. A socioeconomic overview of Georgia's Marine Related Industries and Activities. U.S Depart- 
ment of Commerce, National Oceanic and Atmospheric Administration, National Ocean Service, Silver Spring, Maryland. 
29 pages. 

Friedlander AM, Parrish JD. 1998. Habitat characteristics affecting fish assemblages on a Hawaiian coral reef. Journal of 
Experimental Marine Biology and Ecology 224:1-30. 

Froese R, D Pauly. Editors. 2005. FishBase. World Wide Web electronic publication, www.fishbase.org, version (09/2005). 

Gilligan, MR 1989. An Illustrated Field Guide to the Fishes of Gray's Reef National Marine Sanctuary. NOAA Technical 
Memorandum, NOS MEMD 25. Marine and Estuarine Management Division, OOCRM, NOS, NOAA, U.S. Department of 
Commerce, Washington, D.C. February 1989. 77 p. 

Gilmore, RG, RS Jones. 1992. Color variation and associated behavior in the epinepheline groupers, Mycteroperca microl- 
epis (Goode and Bean) and M.phenax (Jordan and Swain). Bulletin of Marine Science 51 :83-103. 

Gratwicke, B, MR Speight. 2005. Effects of habitat complexity on Caribbean marine fish assemblages. Marine Ecology 
Project Series 292:301-310. 

Grigg, RW. 1994. Effects of sewage discharge, fishing pressure and habitat complexity on coral ecosystems and reef fishes 
in Hawaii. Marine Ecology Progress Series 103:25-34. 

Grimes, CB, CS Manooch, GR Huntsman. 1982. Reef and rock outcropping fishes of the outer continental shelf of North 
Carolina and South Carolina, and ecological notes on the red porgy and vermilion snapper. Bulletin of Marine Science 
32:277-289. 

Harris, PJ, DM Wyanski, DB White, JL Moore. 2002. Age, growth, and reproduction of scamp, Mycteroperca phenax, in the 
Southwestern North Atlantic, 1979-1997. Bulletin of Marine Science 70:113-132. 

Hunt, JL, Jr. 1 974. The geology and origin of Gray's Reef, Georgia continental shelf. M.S. thesis, Univ. Georgia, Athens, 83 
p. 

Huntsman, GR. 1976. Offshore headboat fishing in North Carolina and South Carolina. Marine Fisheries Review. 38:13- 
23. 

Hyland, J, C Cooksey, WL Balthis, M Fulton, D Bearden, G McFall, M Kendall. 2006. The soft-bottom macrobenthos of 
Gray's Reef National Marine Sanctuary and nearby shelf waters off the coast of Georgia, USA. Journal of Experimental 
Marine Biology and Ecology 330:307-326. 

Jennings, S, NVC Polunin. 1996. Effects of fishing effort and catch rates upon the structure and biomass of Fijian reef fish 
communities. Journal of Applied Ecology 33:400-412. 



Jennings, S, NVC Polunin. 1997. Impacts of predator depletion by fishing on the biomass and diversity of non-target reef 
fish communities. Coral Reefs 16:71-82. 

Jennings, S, EM Grandcourt, NVC Polunin. 1995. The effects of fishing on the diversity, biomass and trophic structure of 
Seychelles' reef fish communities. Coral Reefs 14:225-235. 

Kendall, MS, OP Jensen, C Alexander, D Field, G McFall, R Bohne, ME Monaco. 2005. Benthic mapping using sonar, video 
transects, and an innovative approach to accuracy assessment: A characterization of bottom features in the Georgia Bight. 
Journal of Coastal Research 21:1154-1165. 

Luckhurst, BE, K Luckhurst. 1978. Analysis of the influence of substrate variables on coral reef fish communities. Marine 
Biology 49:317-323. 

Matheson, RH III, GR Huntsman, CS Manooch. 1986. Age, growth, mortality, food, and reproduction of the scamp, Mycte- 
roperca phenax, collected off North Carolina and South Carolina. Bulletin of Marine Science 38:300-312. 

McGovern, JC, GR Sedberry, HS Meister, TM Westendorff, DM Wyanski, PJ Harris. 2005. A tag and recapture study of gag, 
Mycteroperca microlepis, off the Southeastern US. Bulletin of Marine Science 76:47-59. 

McClanahan, TR. 1994. Kenyan coral reef lagoon fish: effects of fishing, substrate complexity, and sea urchins. Coral Reefs 
13:231-241. 

Menza, C, J. Ault, J. Beets, J. Bohnsack, C. Caldow, J. Christensen, A. Friedlander, C. Jeffrey, M. Kendall, J. Luo, M. Mo- 
naco, S. Smith, K. Woody. 2006. A Guide to Monitoring Reef Fish in the National Park Service's South Florida / Caribbean 
Network. NOAA Technical Memorandum NOS NCCOS 39. 169 pp. 

Mercer, LP. 1989. Species profiles: Life histories and environmental requirements of coastal fishes and invertebrates (South 
Atlantic) - black sea bass. US Fish and Wildlife Service Biological Report 82(1 1 .99). US Army Corps of Engineers, TR EL- 
82-4. 16 pp. 

Miller, GC, WJ Richards. 1980. Reef fish habitat, faunal assemblages, and factors determining distributions in the South 
Atlantic Bight. Proceedings of the Gulf and Caribbean Fisheries Institute 32:114-130. 

Molles, MC. 1978. Fish species diversity on model and natural reef patches: Experimental insular biogeography. Ecological 
Monographs 48:289-305. 

NOAA. 2006. Gray's Reef National Marine Sanctuary Final Management Plan/Final Environmental Impact Statement. 
NOAA NOS NMSP, Savannah, GA. 

Ohman, MC, A Rajasuriya. 1998. Relationship between habitat structure and fish communities on coral and sandstone 
reefs. Environmental Biology of Fishes 53:19-31. 

Parker Jr., RO. 1990. Tagging studies and diver observations offish populations on live-bottom reefs of the U.S. southeast- 
ern coast. Bulletin of Marine Science 46:749-760. 

Parker Jr., RO, RL Dixon. 1998. Changes in a North Carolina reef fish community after 15 years of intense fishing- global 
warming implications. Transactions of the American Fisheries Society 127:908-920. 

Parker Jr., RO, RW Mays. 1998. Southeastern US deepwater reef fish assemblages, habitat characteristics, catches, and 
life history. US Department of Commerce. NOAA Technical Report NMFS 138. 41 p. 

Parker Jr., RO, SW Ross. 1986. Observing reef fishes from submersibles off North Carolina. Northeast Gulf Science 8:31- 
49. 

Parker Jr., RO,AJ Chester, RS Nelson. 1994. A video transect method for estimating reef fish abundance, composition, and 
habitat utilization at Gray's Reef National Marine Sanctuary, Georgia. Fishery Bulletin 92:787-799. 

Powles, H, CA Barans. 1980. Groundfish monitoring in sponge-coral areas off the Southeastern United States. Marine 
Fisheries Review 42:21-35. 

Quattrini, AM, SW Ross. 2006. Fishes associated with North Carolina shelf-edge hardbottoms and initial assessment of a 
marine protected area. Bulletin of Marine Science 79:137-163. 

Reef Environmental Education Foundation, http://www.reef.org/ 



Riggs, SR, SW Snyder, AC Hine, DL Mearns. 1996. Hardbottom morphology and relationship to the geologic framework: 
mid-Atlantic continental shelf. Journal of Sedimentary Research 66:830-846. 

Risk, MJ. 1972. Fish diversity on a coral reef in the Virgin Islands. Atoll Research Bulletin 153:1-6. 

Roberts, CM, RFG Ormond. 1 987. Habitat complexity and coral reef fish diversity and abundance on Red Sea fringing reefs. 
Marine Ecology Progress Series 41 :1-8. 

Russ, GR, AC Alcala. 1989. Effects of intense fishing pressure on an assemblage of coral reef fishes. Marine Ecology Prog- 
ress Series 56:13-27. 

Russ, GR, AC Alcala. 1998. Natural fishing experiments in marine reserves 1983-1993: community and trophic responses. 
Coral Reefs 17:383-397. 

Sedberry, GR, RF Van Dolah. 1984. Demersal fish associated with hard bottom habitat in the South Atlantic Bight of the 
U.S.A.. Environmental Biology of Fish 11:241-258. 

Sedberry, GR, CL Cooksey, SF Crowe, J Hyland, PC Jutte, CM Ralph, LR Sautter. 2004. Characterization of deep reef 
habitat off the Southeastern US with particular emphasis on discovery, exploration and description of reef fish spawning 
sites. South Carolina Department of Natural Resources, Marine Resources Research Institute. Charleston , SC. Project 
NA16RP2697. 76 p. 

South Atlantic Fishery Management Council. 2006. http:/www.safmc.net/ Accessed September 1 , 2006. 

Struhsacker, P. 1969. Demersal fish resources: Composition, distribution, and commercial potential of the continental shelf 
stocks off southeastern United States. Fisheries Industrial Research 4:261-300. 

Wantiez, L, P Thollot, M Kulbicki. 1997. Effects of marine reserves on coral reef fish communities from five islands in New 
Caledonia. Coral Reefs 16:215-224. 

Watson, M, RFG Ormond. 1994. Effect of an artisanal fishery on the fish and urchin populations of a Kenyan coral reef. 
Marine Ecology Progress Series 109:115-129. 

Wenner, CA. 1983. Species associations and day-night variability of trawl-caught fishes from the inshore sponge-coral habi- 
tat, South Atlantic Bight. Fishery Bulletin 81:537-552. 

Wenner, CA, CA Barans, BW Stender, FH Berry. 1 979a. Results of MARMAP otter trawl investigations in the South Atlantic 
Bight. II. Spring, 1974. Marine Resources Research Institute. South Carolina Wildlife and Marine Resources Department. 
Charleston, South Carolina. Technical Report Number 40. 78 pages. 

Wenner, CA, CA Barans, BW Stender, FH Berry. 1 979b. Results of MARMAP otter trawl investigations in the South Atlantic 
Bight. II. Summer, 1974. Marine Resources Research Institute. South Carolina Wildlife and Marine Resources Department. 
Charleston, South Carolina. Technical Report Number 41. 62 pages. 

Westera, M, P Lavery, G Hyndes. 2003. Differences in recreationally targeted fishes between protected and fished areas of 
a coral reef marine park. Journal of Experimental Marine Biology and Ecology 294:145-168. 



Appendix A. Fish species observed on all bottom types. Within each of the major bottom types at GRNMS, the percent of surveys 
on which the species was encountered and the average abundance and biomass (and standard error) are provided. Presence during 
the May and/or August survey periods is denoted by an X. For species which have zero values for probability of encounter, abun- 
dance and biomass are left blank. No standard error is given when a species was seen on less than three surveys although mean 
abundance and biomass are provided. Also note that mean values are rounded to the ones digit and SE is rounded to tenths which 
results in some low values appearing as zeros. 



Genus species 

common name 

Abudefduf saxatilis 

sergeant major 



May 



Acanthostracion quadricornis X 

scrawled cowfish 



Antennarius sp. 



frogfish 



Apogon pseudomaculatus X 

twospot cardinalfish 

Archosargus probatocephalus X 
sheepshead 



Archosargus rhomboidalis X 

sea bream 



Balistes capriscus X 

gray triggerfish 



Bothus ocellatus X 

eyed flounder 

Calamus bajonado X 

jolthead porgy 

Calamus calamus X 

saucereye porgy 

Calamus penna X 

sheepshead porgy 



Caranx bartholomaei 



yellow jack 



Caranx crysos X 

blue runner 



Caranx ruber 



bar jack 



Centropristis ocyurus X 

bank sea bass 

Centropristis striata X 

black sea bass 

Chaetodipterus faber X 

Atlantic spadefish 

Chaetodon ocellatus 

spotfin butterflyfish 

Chilomycterus schoepfi X 

stripped burrfish 

Chloroscombrus chrysurus 

Atlantic bumper 



Aug 

X 



Variable 

percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 









Sparse live 






Flat sand 


Rippled sand 


bottom 


Ledge 




Value SE 


Value SE 




Value SE 




Value SE 







1 
























1 











3 













I 












° 







49 










3 0.5 












6 1.1 







33 










2 0.7 






6 


25 


3041 839.7 




5 


20 




1 





2 0.5 


1 0.3 




138 


10 



139 47.8 



652 507.7 







26 










1 0.3 






13 





846 290.6 




20 







0.1 











0.2 



















3 










0.0 












176 120.5 







10 










0.1 












229 98.5 







5 










0.1 










49 33.0 










8 

0.1 ■ 


3 

0.0 






6 


13 6.9 1 
22 


15 13.2 


1 


15 


24 


14 10.2 





5 2.4 


21 13.4 


7114 5322.2 


5 


6 


3026 1560.2 
33 


14278 9082.1 







1 0.6 
290 200.5 




5 


41 










1 0.5 1 


1 0.3 




1 



6 


65 34.4 
98 


106 29.5 




5 


98 










13 1.5 1 


28 2.3 




9 


38 



1327 193.3 1 
2 


4111 524.0 







12 








1 


7 3.4 









182 



3070 1572.9 







2 







































13 


12 







11 


12 




2 





8 4.5 1 


24 12.4 




146 


8 


578 289.2 1 


5176 3023.0 



Appendix A. Continued. Fish species observed on all bottom types. Within each of the major bottom types at GRNMS, the percent 
of surveys on which the species was encountered and the average abundance and biomass (and standard error) are provided. 
Presence during the May and/or August survey periods is denoted by an X. For species which have zero values for probability of 
encounter, abundance and biomass are left blank. No standard error is given when a species was seen on less than three surveys 
although mean abundance and biomass are provided. Also note that mean values are rounded to the ones digit and SE is rounded 
to tenths which results in some low values appearing as zeros. 



Genus species 



common name 



May 



Conger sp. 



conger eel 



Coryphopterus glaucofraenum 
bridled goby 



Decapterus sp. 



Diodon hystrix 



scad 



porcupinefish 



Diplectrum formosum 

sand perch 



Diplodus holbrookii 

spottail pinfish 

Echeneis naucrates 

sharksucker 



Epinephelus morio 

red grouper 

Equetus lanceolatus 

jackknife fish 

Ginglymostoma cirratum 

nurse shark 

Gymnachirus melas 

naked sole 

Gymnothorax saxicola 

honeycomb moray 



" 



aemulon aurolineatum 



tomtate 



Haemulon plumierii 



Haemulon sp. 



white grunt 



grunt 



Halichoeres bivittatus 

slippery dick 

Halichoeres caudalis 

painted wrasse 

Holacanthus bermudensis 

blue angelfish 

Hypleurochilus geminatus 

crested blenny 



Lutjanus analis 



Aug 

X 



mutton snapper 



Variable 

percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 



Flat sand 
Value SE 








20 

14 

1814 





11 




12.6 
1776.6 



Rippled sand 
Value SE 





38 
8 3.9 

628 524.8 




38 

1 0.3 

47 42.8 





Sparse live 

bottom 
Value SE 



1 

240 





43 
4 

71 

6.00 



30 
2 






22 







25 

4 





0.5 
205.4 






0.9 
21.4 

0.4 1 
26.7 ■ 



0.1 

15.3 



0.0 
14.6 




Ledge 
Value 



SE 



1 




o 




3 


■ 


5 




o 


0.0 


° 


0.1 


10 




195 


118.7 


908 


540.5 


1.00 




o 




19 




33 




1 


0.4 


57 


22.3 


34.00 




4 


1.0 


483 


112.4 


' 


7.00 




o 


0.0 


102 


43.7 


16 







0.1 


26 


10.9 


3 







0.0 


1006 


776.9 





1 














48 




931 


494.5 


1897 


644.2 


11 







0.2 


240 


153.8 


i 


89 




15 


1.6 


290 


37.0 


45 




1 


0.2 


57 


12.6 


20 




1 


0.2 


603 


180.5 


17 




o 


0.1 





0.1 



12 



Appendix A. Continued. Fish species observed on all bottom types. Within each of the major bottom types at GRNMS, the percent 
of surveys on which the species was encountered and the average abundance and biomass (and standard error) are provided. 
Presence during the May and/or August survey periods is denoted by an X. For species which have zero values for probability of 
encounter, abundance and biomass are left blank. No standard error is given when a species was seen on less than three surveys 
although mean abundance and biomass are provided. Also note that mean values are rounded to the ones digit and SE is rounded 
to tenths which results in some low values appearing as zeros. 



Genus species 

common name 

May 

Lutjanus campechanus X 

red snapper 

Microgobius earn X 

seminole goby 



Micropogonias undulatus 

Atlantic croaker 

Muraena retifera X 

reticulate moray 

Mycteroperca microlepis X 

gag grouper 

Mycteroperca phenax X 

scamp 

Nicholsina usta 

emerald parrotfish 



Ogcocephalus nasutus 

shortnose batfish 

Ogcocephalus radiatus X 

polka-dot batfish 

Opsanus tau X 

oyster toadfish 

Pagrus pagrus X 

red porgy 

Parablennius marmoreus X 

seaweed blenny 



" 



aralichthys albigutta X 

gulf flounder 



Pareques sp. X 

cubbyu/high hat 

Pomacanthus paru 

French angelfish 



Pomacanthus sp. 



angelfish 



Prionotus ophryas X 

bandtail searobin 

Prionotus scitulus X 

leopard searobin 



Prionotus sp. 



searobin 



Ptereleotris calliurus X 

blue goby 



Aug 

X 



Variable 

percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 



I 



Flat sand 
Value SE 





20 
0.2 

0.2 



I 



10 



25 
0.1 

3 2.2 



Rippled sand 
Value SE 





31 
1 
1 




0.3 
0.3 



Sparse live 

bottom 
Value 



0.1 
256.9 




56 



17.7 



27.1 



45 











28 
1 0.2 


6 






1 0.3 


8 


1 


16 








0.0 


0.2 


2 




31 


21.0 


309 150.3 


2 




55 









55 22.8 










6013 3411.1 





: 










1 













2 



















4 











4 








0.0 


6 






6 3.0 


12 




11 








0.0 


0.0 


2 




22 


10.8 


8 3.3 


10 




5 







0.1 1 


0.1 




1 


1.0 | 


0.3 



Appendix A. Continued. Fish species observed on all bottom types. Within each of the major bottom types at GRNMS, the percent 
of surveys on which the species was encountered and the average abundance and biomass (and standard error) are provided. 
Presence during the May and/or August survey periods is denoted by an X. For species which have zero values for probability of 
encounter, abundance and biomass are left blank. No standard error is given when a species was seen on less than three surveys 
although mean abundance and biomass are provided. Also note that mean values are rounded to the ones digit and SE is rounded 
to tenths which results in some low values appearing as zeros. 



Genus species 

common name 

Ptereleotris helenae 

hovering goby 



May 



Raja eglanteria X 

clearnose skate 

Rhinobatos lentiginosus X 

Atlantic guitarfish 

Rypticus maculatus X 

whitespotted soapfish 

Scomberomorus maculatus 

Spanish mackerel 

Scorpaena sp. 

scorpionfish 

Seriola sp. 

almaco/amberjack 

Serraniculus pumilio 

pygmy sea bass 

Serranus subligarius X 

belted sandfish 

Sphyraena barracuda X 

great barracuda 

Sphyraena sp. 



Stegastes variabilis 

cocoa variabilis 



Stenotomus sp. X 

scup/longspine porgy 

Stephanolepis hispidus X 

planehead filefish 

Syngnathidae sp. X 

pipefish 

Synodus sp. X 

lizardfish 

Urophycis earlli X 

Carolina hake 

Xyrichtys novacula X 

pearly razorfish 



Aug 

X 



Variable 

percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 



I 



Flat sand 
Value SE 



I 

I 



10 
3539 



35 



15 



26 





7.5 
18.5 



15.00 

0.1 

23 17.9 



75 

3 0.7 

44 12.2 



Rippled sand 
Value SE 





31 

69 62.4 
257 161.6 




13 





6.00 



24 




Sparse live 

bottom 
Value SE 





9 
68 



6.8 
19.5 



2 



48 



2 

1 
4 
2 
537 




4 



149 





47 
2 
6 




71 
4.00 




0.6 
3.4 



90 

20 4.1 

1677 313.0 





54.4 






10 

1 



8.00 


14 
1 

12.00 2.00 

0.1 I 



26.00 

1 

283 

T: 



0.4 
71.8 



11 

39 

1 

167 

7 

3 1.7 

970 599.0 

1 



1 

9 



276 

1 





88 

13 

25 

7 



162 

1 



0.2 
198.7 



1.4 
2.8 



0.1 
81.5 



0.2 
0.6 



24 2.8 

3007 430.0 



0.0 
5.5 




0.4 
79.4 



Appendix B. Fish species observed on ledges. Within each of the four ledge types based on cluster analysis, the percent of surveys 
on which the species was encountered and the average abundance and biomass (and standard error) are provided. For species 
which have zero values for probability of encounter, abundance and biomass are left blank. No standard error is given when a species 
was seen on less than three surveys although mean abundance and biomass are provided. Also note that mean values are rounded 
to the ones digit and SE is rounded to tenths which results in some low values appearing as zero's. 



Genus species 

common name 

Abudefduf saxatilis 

sergeant major 

Acanthostracion quadricornis 
scrawled cowfish 



Antennarius sp. 



frogfish 



Apogon pseudomaculatus 

twospot cardinalfish 

Archosargus probatocephalus 
sheepshead 

Archosargus rhomboidalis 

sea bream 

Balistes capriscus 

gray triggerfish 

Bothus ocellatus 

eyed flounder 

Calamus bajonado 

jolthead porgy 

Calamus calamus 

saucereye porgy 

Calamus penna 

sheepshead porgy 

Caranx bartholomaei 

yellow jack 



Caranx crysos 



blue runner 



Caranx ruber 



bar jack 



Centropristis ocyurus 

bank sea bass 

Centropristis striata 

black sea bass 

Chaetodipterus faber 

Atlantic spadefish 

Chaetodon ocellatus 

spotfin butterflyfish 

Chilomycterus schoepfi 

stripped burrfish 

Chloroscombrus chrysurus 

Atlantic bumper 



Variable 

percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 




Cluster 2 
Value 



SE 






4 



43 

17 



313 

9 



35 

4 



52 

26 

58 

39360 

22 

4 

1152 

39 

1 

107 

96 

35 

6702 

17 

12 

4667 

4 









9 

44 

10923 



0.1 

180.2 



0.2 

52.3 

52.0 
34944.5 

2.5 

787.5 



14 2.5 ■ 

1152 787.5 ■ 

39 I 

1 0.4 

I 



0.4 
58.2 1 

6.8 
1620.6 



17 




o 








• 


M 


67 




5 


1.7 


11 


3.9 


83 




6 


2.1 


7117 


3380.8 


" 


67 




2 


0.7 


952 


385.6 





33 




1 


0.5 


2527 


1684.6 


33 




1 


0.8 


1794 


1221.7 


17 









492 







50 




62 


48.2 


41552 


36637.3 


17 




1 




24 




33 


m 


1 


0.6 


145 


141.9 


100 




30 


11.7 


5368 


3228.7 


83 




56 


34.0 


26331 


18472.4 








Cluster 3 
Value 



SE 












31 




1 


0.7 


3 


1.5 


8 







0.3 


593 


333.7 


20 




1 


0.3 


123 


77.5 


6 







0.1 


123 


87.6 








2 









25 




4 


■■ 





0.1 


14 


10.1 


2 









3 




22 


^H 


5 


1.9 


2824 


1335.5 


' 


33 




2 


0.4 


96 


40.5 


98 




23 


1.9 


2879 


320.0 


4 




2 


1.4 


337 


236.5 















Cluster 4 
Value SE 









83 

2 

123 

100 

35 

3747 





1.7 
1243.2 



0.6 
75.7 

5.2 
1077.3 



43.4 
10700.9 



83.3 

20524.7 



5.2 
1059.6 




Appendix B. Continued. Fish species observed on ledges. Within each of the four ledge types based on cluster analysis, the per- 
cent of surveys on which the species was encountered and the average abundance and biomass (and standard error) are provided. 
For species which have zero values for probability of encounter, abundance and biomass are left blank. No standard error is given 
when a species was seen on less than three surveys although mean abundance and biomass are provided. Also note that mean 
values are rounded to the ones digit and SE is rounded to tenths which results in some low values appearing as zero's. 



Genus species 

common name 



Conger sp. 



conger eel 



Coryphopterus glaucofraenum 
bridled goby 



Decapterus sp. 
Diodon hystrix 



scad 



porcupinefish 

Diplectrum formosum 

sand perch 

Diplodus holbrookii 

spottail pinfish 

Echeneis naucrates 

sharksucker 



Epinephelus mono 

red grouper 

Equetus lanceolatus 

jackknife fish 

Ginglymostoma cirratum 

nurse shark 

Gymnachirus melas 

naked sole 

Gymnothorax saxicola 

honeycomb moray 

Haemulon aurolineatum 

tomtate 



Haemulon plumierii 



Haemulon sp. 



white grunt 



grunt 



Halichoeres bivittatus 

slippery dick 

Halichoeres caudalis 

painted wrasse 

Holacanthus bermudensis 

blue angelfish 

Hypleurochilus geminatus 

crested blenny 

Lutjanus analis 

mutton snapper 



Variable 

percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 



Cluster 4 
Value SE 




Appendix B. Continued. Fish species observed on ledges. Within each of the four ledge types based on cluster analysis, the per- 
cent of surveys on which the species was encountered and the average abundance and biomass (and standard error) are provided. 
For species which have zero values for probability of encounter, abundance and biomass are left blank. No standard error is given 
when a species was seen on less than three surveys although mean abundance and biomass are provided. Also note that mean 



Genus species 

common name 

Lutjanus campechanus 

red snapper 

Microgobius cam 

seminole goby 

Micropogonias undulatus 

Atlantic croaker 

Muraena retifera 

reticulate moray 

Mycteroperca microlepis 

gag grouper 

Mycteroperca phenax 

scamp 



M 



'icholsina usta 

emerald parrotfish 



Ogcocephalus nasutus 

shortnose batfish 

Ogcocephalus radiatus 

polka-dot batfish 



Opsanus tau 



oyster toadfish 



Pagrus pagrus 



red porgy 



Parablennius marmoreus 

seaweed blenny 

Paralichthys albigutta 

gulf flounder 

Pareques sp. 

cubbyu/high hat 

Pomacanthus paru 

French angelfish 



Pomacanthus sp. 



angelfish 



Prionotus ophryas 

bandtail searobin 

Prionotus scitulus 

leopard searobin 



Prionotus sp. 



searobin 



Ptereleotris calliurus 



blue goby 



Variable 

percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 



I 






3ea to ic 


;ntns wnic 
er1 

se| 


n results in some ic 

Cluster 2 
Value SEl 

33 
2 1.1 
4219 3450.5 



>w values appearing 

Cluster 3 
Value SE 



8 

0.1 

1 1.0 


as zero s. 

Cluster 4 
Value SE 


Q_ 
Q. 


Clust 
Value 


< 


22 



793 


0.2 
432.3 


8 








4 




0.2 
0.2 1 


50 
1 
1 


0.3 
0.3 




: 1 


mm 

50 

1 

132 

83 
5 1.3 
13386 3824.1 
100 
11 3.3 
16701 6939.8 





2 


10 

4 

0.1 
689 670.3 

2 


12 







■ 




17 



101 


0.1 1 

54.2 yj 


17 



28 






43 

1 

5283 


0.7 
3700.8 


8 



79 


■ 




61 

4 

6141 


1.1 1 

1997.9 1 


8 

1 

3095 


1.3 
3095.3 




















100 

3 0.8 
504 140.9 



67 

3 1.2 
5 2.0 

50 

4 2.1 
2338 2145.3 

100 

190 73.8 
13366 3123.5 





2 



1 

2 



6 
55 

1 0.1 
141 30.6 

2 

0.2 
75 75.4 
14 

0.1 1 

0.2 [ 

4 


30 
25 

6 3.1 1 
492 359.61 

2 





















61 

1 

215 


0.3 
55.3 


67 

1 

178 


0.3 
82.1 




4 



14 





■ 




52 
2 
3 


0.5 
0.8 


25 
1 
1 


0.4 
0.7 




35 

1 

518 


0.3 
194.6 


17 



81 


■ 




96 

154 

19444 


85.3 
13361.6 


83 

8 

61 


1.7 
50.4 














4 

1 

























8 

0.0 1 
1 10 5.4 [ 

16 

0.1 
11 5.0 1 

2 

0.1 1 

0.1 1 











7 






8 




■ 









33 

I 


0.2 
2.2 






Appendix B. Continued. Fish species observed on ledges. Within each of the four ledge types based on cluster analysis, the per- 
cent of surveys on which the species was encountered and the average abundance and biomass (and standard error) are provided. 
For species which have zero values for probability of encounter, abundance and biomass are left blank. No standard error is given 
when a species was seen on less than three surveys although mean abundance and biomass are provided. Also note that mean 
values are rounded to the ones digit and SE is rounded to tenths which results in some low values appearing as zero's. 



Genus species 

common name 

Ptereleotris helenae 

hovering goby 

Raja eglanteria 

clearnose skate 



Rhinobatos lentiginosus 

Atlantic guitarfish 

Rypticus maculatus 

whitespotted soapfish 

Scomberomorus maculatus 
Spanish mackerel 

Scorpaena sp. 

scorpionfish 

Seriola sp. 

almaco/amberjack 

Serraniculus pumilio 

pygmy sea bass 

Serranus subligarius 

belted sandfish 

Sphyraena barracuda 

great barracuda 



Sphyraena sp. 



barracuda 



Stegastes variabilis 

cocoa variabilis 

Stenotomus sp. 

scup/longspine porgy 

Stephanolepis hispidus 

planehead filefish 



Syngnathidae sp. 
Synodus sp. 



Urophycis earlli 



pipefish 



lizardfish 



Carolina hake 

Xyrichtys novacula 

pearly razorfish 



Variable 

percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 
percent of surveys 
mean abundance 
mean biomass (g) 



I 



Cluster 1 
Value SE 





Cluster 2 
Value 





SE 



4.3 
995.1 




Cluster 3 


Value 


SE 


2 


















2 









19 




16 







0.1 


21 


8.8 


2 







0.1 


42 


41.6 





2 









5 




2 














78 




7 


1.2 


15 


3.3 


4 









61 







8 







0.1 





0.3 


98 




32 


3.8 


3470 


454.4 


6 









13 




2 














4 









17 




31 




2 


0.6 


383 


130.3 



Cluster 4 


Value 


SE 








• 


58 




1 


0.3 


64 


40.1 





8 









6 




17 







0.3 


62 


47.0 





100 




23 


5.7 


39 


9.6 


8 









370 







8 




1 


1.2 


4 


3.6 


92 




31 


6.6 


4218 


1465.1 


8 









5 










8 









19 





ACKNOWLEDGEMENTS 



Chris Caldow, Randy Clark, Kelly Gleason, Jenny Waddell, and Kim Woody assisted in the collection of the 
field data at Gray's Reef. Thanks to the boat captains Keith Golden, Jay Fripp, and Todd Recicar as well as 
the officers and crew of the NOAA ship Nancy Foster for safely getting us to and from our field sites. We are 
grateful to Captain Judy Helmey, William H. "Bing" Phillips, and John Duren for valuable discussions regarding 
recreational fishing at Gray's Reef. Thanks to Randy Clark, Charlie Menza, Simon Pittman, Jenny Waddell, 
and Adam Zitello of the Biogeography Team for providing helpful comments and editorial assistance. Sarah 
Davidson-Hile organized and improved the many photos taken during the field missions. Jamie Higgins format- 
ted and organized the figures, tables, images, and text into the work of art before you. Thanks to the GRNMS 
staff, particularly Greg McFall for fielding questions throughout the process and for helpful suggestions to 
improve this document. Funding for this study was provided by GRNMS and the NMSP/NCCOS Long Term 
Agreement. 





< 
at 

73 

CD 




CD O 






harac 
f Nati 




O i^ 






3 CD 




0) 2. 
— N 




ation of 
Marine 




CO ~ 

0) 3" 




3 CD 




a ro 






enthos, Ma 
uary - NOA 




trine 

ATe 






Debris and Bo 

chnical Memora 




1? 




i 3 

z 3! 

O » 

C/D J 

z: ** 

o 
o 
o 

CO 

Ol 

o 









United States Department of Commerce 

Carlos M. Gutierrez 
Secretary 

National Oceanic and Atmospheric Administration 

Vice Admiral Conrad C. Lautenbacher, Jr. USN (Ret.) 
Under Secretary of Commerce for Oceans and Atmopsheres 

National Ocean Service 

John H. Dunnigan 
Assistant Administrator 


gg^jj F|5 


,-—-- —- 


\^ of Science, Service, & Stewardship ^/ 


v>^M w^