86 
Fishery Bulletin 1 13(1) 
Digital still photographs were taken at 1-min inter- 
vals along each transect for a minimum of 20 photo- 
graphs. Each still photograph covered an area of ap- 
proximately 0.42 m 2 . Paired, parallel, down-looking 
lasers (spaced 10 cm apart on the ROV) provided a con- 
sistent reference for still photographs to maintain con- 
stancy in area of coverage for each image and to size 
individual organisms where possible. Still photographs 
were used to assess the primary metrics of the study, 
namely 1) the percent cover of microtopographic fea- 
tures, meaning the proportion of the seafloor that had 
an elevation, burrow, or other microtopographic feature, 
2) density of structure-forming macro-invertebrates, 
and 3) density of selected mobile macro-invertebrates. 
From our observations across multiple ROV surveys, 
we determined that the microtopographic complexity 
in the unconsolidated sandy sediments at the study 
depths primarily resulted from bioturbation. Bioturba- 
tion in this context refers to deviations from the plane 
of the sediment-water interface (such as ridges and 
mounds, burrows and hole) created by the movement 
of organisms such as sea stars and fishes or organisms 
such as mud urchins through the upper centimeters of 
the sediment. Small features that result from biotur- 
bation can serve as habitat for demersal fishes from 
a variety of species, including many flatfishes found 
in the study area. Digital still photographs were used 
to quantify the spatial extent of microtopographic fea- 
tures (i.e., bioturbation) in each of the 8 study plots. 
The percent area covered was quantified for each still 
photograph with a digitally rendered grid that was 5 
by 5 cm and was superimposed over each photograph. 
Any cell in which a microtopographic feature was evi- 
dent was counted. 
Digital still photographs also were used to quan- 
tify the density of benthic invertebrate species in 
each study plot. Structure-forming organisms (such 
as those found in the study area, primarily sea pens, 
sea whips, and anemones) are erect, mostly sedentary 
creatures that extend above the plane of the sediment 
and provide potential habitat structure for fishes and 
mobile invertebrates. Mobile organisms, including 
crabs, urchins, and sea stars, move along and across 
the plane of the sediment, providing transient habitat 
themselves (e.g., for small fish seeking cover adjacent 
to larger sea stars) and altering the sediment itself, 
as described previously. Densities of both structure- 
forming and mobile invertebrates were extracted from 
digital still photographs. Every organism that occurred 
in still photographs was counted and identified to the 
lowest taxonomic level possible and then grouped by 
taxon or mobility for analyses. 
Data analyses 
Although a BACI sampling design was planned (similar 
to the one used by Pitcher et ah, 2009) with 8 random- 
ly distributed plots, the limitations, described previ- 
ously, in the distribution of trawling effort across those 
plots required that we alter our analytical approach. 
To optimize our ability to detect statistical differences 
among paired plots and across years and to reduce the 
probability of a type-II error, we randomly selected 
50 photographs from each of 4 pairs of adjacent plots 
for each treatment (identified as analysis units in Fig. 
1) so that independent samples would be distributed 
across the entire study area and across time. This prac- 
tice effectively reduced the number of replicates for 
each treatment from 4 plots to 2 pairs of adjacent plots 
and allowed for a greatly improved number of samples 
(A0, when compared with a transect-based approach: 
A=1600 for treatment-wide analyses, and A=200 for 
year-to-year comparisons. This approach also reduced 
the effect of spatial autocorrelation in data collected 
along sequential transects and eliminated any inflated 
statistical significance due to psuedoreplication (Hurl- 
bert, 1984). 
We used the 2-proportion Z-test to determine wheth- 
er the hypothesized difference between population pro- 
portions (arcsine transformed) in microtopographic 
structure varied significantly between trawled and con- 
trol plots and within treatments (high-intensity or low- 
intensity trawling effort) over time. T-tests were used 
to test for any differences in the densities of structure- 
forming macroinvertebrates and mobile macroinverte- 
brates between trawled and control plots and within 
treatments over time. Post-hoc power analyses were 
conducted with the statistical program R, vers. 3.1.0 
(R Core Team, 2014) on the Z-test and t-tests for the 3 
primary metrics to evaluate the power of each test to 
detect significant differences, particularly given the low 
densities observed for sessile, structure-forming inver- 
tebrates and for mobile invertebrates. 
Results 
From September 2009 to May 2012, ROV surveys were 
completed along 297 transects in the study area, and 
6200 still photographs were taken on transects (an 
additional 11,000 photographs were taken to support 
species identifications and to capture unique features). 
From those still photographs, we extracted data on 38 
invertebrate species from 5 phyla — anthozoans (10), 
echinoderms (11), molluscs (11), crustaceans (3), and 
annelids (3). Polychaete worms ( Chloeia pinnata ) were 
numerically dominant, followed closely by several spe- 
cies of brittle stars (phylum Echinodermata: class 
Ophiuroidea), all of which occurred on the surface of 
the sediment in dense but spatially and temporally 
variable patches. 
Temporal and spatial variability 
Although overall density of these observed species was 
low (fractions of individuals per square meter) in the 
study area, temporal variability in the densities of se- 
lected invertebrate taxa was considerable across sam- 
