Harter et al.: Fish assemblages associated with grouper pits in the Gulf of Mexico 
423 
B 
O 2013 
II 2014 
☆ 2015 
□ Central basin 
☆ West ridge 
Figure 2 
Locations of each grouper pit surveyed during 2012-2015 in the Gulf of Mex- 
ico, characterized by 3 of the factors used to compare fish assemblages at 
pits: (A) year, (B) predator presence, and (C) region. The polygon outlined in 
black represents the Pulley Ridge Habitat Area of Particular Concern, which 
is located off southwestern Florida. Predator presence or absence categories 
were no predator, red grouper ( Epinephelus morio) only, species of lionfish 
(Pterois spp.) only, or both predators. 
that “down-weighs” the dominance of highly abundant 
species, allowing species with intermediate density to 
exert some influence on the calculation of similarity 
(Clarke and Warwick, 2001). 
An MDS routine based on Bray-Curtis similarity co- 
efficients and a dendrogram with group-average link- 
ing were created to depict the results of a concurrent 
similarities profile (SIMPROF routine in PRIMER). 
The MDS routine is an ordination technique in which 
points that are located closer together in multivariate 
space are considered more similar than points further 
away. The stress values shown in MDS plots reflect the 
accuracy of the representation of community structure; 
lower stress values indicate that the plots are increas- 
ingly representative of the commu- 
nity structure. Stress values less 
than 0.20 generally indicate that 
plots provide an accurate represen- 
tation of the data rather than that 
the points have been placed arbi- 
trarily in the 2-dimensional ordina- 
tion space. 
We also tested for differences 
in community structure among 
pits with one-way ANOSIM tests 
based on the Bray-Curtis similar- 
ity coefficients. Significant differ- 
ences among groups were defined 
in our study when P< 0.05, but for 
those pairwise tests that showed 
significant difference, we further 
examined the ANOSIM i?-statistic. 
Unlike the P-value, the i?-statistic 
reflects the absolute difference in 
community structure between treat- 
ments (i.e., it reflects the size of the 
effect) (Clarke et al., 2006). The 
P-statistic typically ranges from 0 
to 1, with values closer to 1 repre- 
senting more significant separation 
among groups and values closer to 
0 representing no difference among 
groups. It is possible to obtain a sig- 
nificant P-value with an P-statistic 
that is very low when there are 
many replicates at each site, and ob- 
taining a significant P-value in such 
a case would indicate little biologi- 
cally significant separation among 
groups (Clarke, 1993). Negative R- 
values denote unusual situations 
where replicates among groups are 
more similar than within a group. 
Analysis of similarity percentages 
was then used to determine which 
species contributed to the dissimi- 
larities between the group pairs. 
Biodiversity indices derived with 
the DIVERSE routine were com- 
pared among grouper pits for each 
factor. Parameters examined included total number, di- 
versity, and evenness of species in the community. The 
Shannon- Weiner function ( H ’) was used to estimate pit 
diversity as log(pj), where is the proportion of 
the total count arising from the ith species. Pielou’s 
evenness was estimated as ifVlog(S), where S is the 
total number of taxa at a pit (Pielou, 1977). 
Analyses of univariate fish abundance 
Fish associated with grouper pits were divided into 
3 categories: small fish, schooling fish, and large fish. 
This classification was needed because several of the 
species (primarily the economically important species) 
