198 
Fishery Bulletin 107(2) 
so that when they were projecting out at a distance of 
5 m, two of the lasers overlapped. The third laser was 
spaced 10 cm apart from the two overlapping lasers, 
which allowed measurements to be made. This was ini- 
tially used to train the eye to determine the distance 
at which fishes could be identified. Distance was then 
estimated on subsequent dives in 2005. Transect area 
(TA) was then calculated as: 
TA = (LIT) - V 2 ( WD ) (Koenig et al., 2005). (2) 
Mean TA was 372.9 m 2 ±1.8 m 2 . Density of all observed 
fish species was calculated for each transect in 2003 and 
2005. Initial analyses demonstrated that no statistical 
differences were evident between years, so data from 
both years were combined for all analyses. 
Multivariate ecological analyses were conducted using 
PRIMER 5.0 (Primer-E Ltd, Plymouth, U.K.) to exam- 
ine fish assemblage composition among habitat types 
and management areas. A non-metric multi-dimensional 
scaling (MDS) ordination of ROV transects was con- 
structed from a Bray-Curtis similarity matrix of square 
root transformed fish densities. A square root trans- 
formation was used to reduce the disparity between 
uncommon and abundant species by downweighting 
abundant species relative to uncommon species (Clarke, 
1993). Prior to analyses, transects in which no fishes 
were observed were deleted, as the same reason may not 
apply to why two samples are devoid of species. Species 
comprising <0.01% of the total abundance of fish were 
also removed to minimize rare species confounding 
the cluster analysis. All pelagic species were removed 
from PRIMER analyses because we wanted to focus 
on benthic fish species associated with reef habitat. A 
two-way crossed analysis of similarity (ANOSIM) and 
pairwise comparisons were used to detect significant 
differences in fish assemblages among habitat types 
and management areas. 
PRIMER was also used to examine biodiversity among 
habitats and management areas by calculating average 
taxonomic distinctness ( A + ). This statistic uses the taxo- 
nomic distance between every pair of species in a given 
assemblage as the basis for determining relative diver- 
sity (Clarke and Warwick, 1998). Unlike conventional 
diversity indices such as the Shannon-Weiner Index, 
A + is independent of sampling effort. To calculate A + , a 
total list of species observed from ROV transects was 
used. The following taxonomic categories were utilized: 
species, genus, family, order, class, and phylum. Each 
of these represents a node in determining taxonomic 
distances between species pairs. This list along with 
fish density data were used to run a TAXDTEST which 
produces funnel plots where A + is plotted in comparison 
with the mean and 95% confidence limits. 
Densities of grouper were singled out for analysis be- 
cause their declining abundances led the South Atlantic 
Fishery Management Council to establish the OECA. A 
generalized linear model (GLM) (Minitab 13.32, State 
College, PA) was used to test for significant differences 
in grouper densities among management areas and 
habitat types. Individual species of grouper were not 
abundant enough to analyze separately, so all grouper 
species were combined. One-way analysis of variance 
(ANOVA) was used to test for significant differences in 
grouper densities among management areas within each 
habitat type. A significance level of P < 0.05 was applied 
to all analyses, and log transformations were applied 
to correct for unequal variances. Pairwise comparisons 
were performed using Tukey’s honestly significant dif- 
ferences (HSD). 
Habitat quantification analyses 
A digital still image of the seafloor (taken pointing 
straight down from the ROV, perpendicular to the sea- 
floor) was taken every 1-3 min during ROV transects to 
quantify habitat type among management areas. These 
images were imported into an image analysis program 
written at the University of North Carolina-Wilmington, 
emulating the area/length analysis tool of Coral Point 
Count software (CPCe, Dania Beach, FL) (Kohler and 
Gill, 2006). Within each image, a polygon was drawn 
around each distinctive hardbottom area and a habitat 
type assigned to it. Habitat types were the same as those 
used for video analyses with the addition of human arti- 
facts (e.g., fishing line, bottles) and shadow, where all or 
part of an image was blurred, usually from sand being 
stirred up by the ROV. The program then calculates 
the percentage of each habitat type within an image 
based on the number of pixels in each polygon. The 
area of each habitat type was calculated using paired 
lasers (set at a known distance of 10 cm apart) on each 
image. Mean area of still images was 1.2 m 2 ±0.05 m 2 . 
One-way ANOVAs were then used to test for significant 
differences in habitat type percentages among manage- 
ment areas. 
Results 
Fish assessment 
Forty-two ROV dives (65 hours of video footage) were 
completed in 2003 and 2005, resulting in 512 hard- 
bottom 50-m transects: 236 in the OECA, 184 in the 
OHAPC, and 92 in the open area. Among habitat types, 
72 transects were in pavement, 186 in rubble, 210 in 
rock outcrops, 11 in standing dead O. varicosa, and 
33 in live O. varicosa. A total of 62 fish species were 
observed (Table 1). The previously unexplored bioherms 
discovered outside the OHAPC between the two satellite 
areas turned out to be comprised mostly of coral rubble, 
therefore, even though some live and standing dead O. 
varicosa were observed in the open areas, there wasn’t 
enough of it to produce any 50-m transects to be used 
in the analyses. No fish species were exclusive to O. 
varicosa coral (live or standing dead). No grouper spe- 
cies were found on pavement except scamp ( Mycteroperca 
phenax), the most abundant grouper. Tattlers (Serranus 
phoebe ), one of the most abundant small sea basses were 
