Hannah et at: Effects of shrimp trawling on macroinvertebrate abundance and diversity 
33 
Table 1 
Depth range (m), number of transects surveyed, and area (ha) and linear distance (km) of transects surveyed with remotely oper- 
ated vehicle at the four mud seafloor sampling sites at Nehalem Bank, Oregon (see Fig. 1), June 2007. 
Sampling site 
Depth range (m) 
Number of transects 
Area surveyed (ha) 
Distance surveyed (km) 
1A 
148.3-154.4 
5 
1.51 
9.25 
IB 
150.5-153.8 
5 
1.23 
9.30 
2A 
149.4-155.7 
5 
1.16 
9.30 
2B 
146.2-156.0 
6 
0.76 
7.11 
practical limits of visibility. As organisms crossed the 
50% line, they were counted and visually identified to 
the lowest taxonomic level possible, with the exception 
of pandalid shrimp, which were very numerous but were 
not counted. Large white sea whips were all recorded as 
Halipteris spp.; however, some of the smaller specimens 
may actually have been Stylatula spp. Hagfish burrows 
were too numerous to count completely throughout the 
video footage and therefore were subsampled by ran- 
domly selecting three two-minute segments per video- 
tape (26.2-min average duration) and counting all of the 
burrows in these subsamples. 
We estimated the area of the belt transects surveyed 
by combining information on the change in position of 
the ROV over time and the width of the camera field 
of view at the 50% line. The on-ground width of the 
camera’s field of view was calculated by measuring 
the distance between the two laser dots (mm) on a 
standard video screen every 30 seconds (based on the 
imprinted video time stamp) and applying an estab- 
lished relationship between the screen laser width and 
the on-ground width of view of the camera. Position 
data for the ROV were smoothed with a 9-point mov- 
ing average before the distance covered in each 30- 
second interval was measured. A simple straight line 
“distance traveled” was calculated for each 30-second 
interval based on the smoothed position data at the 
start and end points of the interval. This distance was 
then multiplied by the average of the two estimates of 
the width of view at the start and end of each 30-sec- 
ond interval to approximate the total area surveyed. 
These areal estimates were then summed across all 
reviewed 30-second intervals to estimate the area in 
each belt transect. 
To determine how ocean shrimp trawling may have 
affected the size distribution of sea whips, we measured 
the approximate size of all sea whips as they were enu- 
merated at the 50% line. The measurement was taken 
when the base of the organism was at the 50% line, 
expressed as a multiple of the laser width separation 
(10 cm) in that video frame. Frequently, with the base 
of large sea whips positioned at the 50% line, only the 
bottom portion of the sea whip was visible. In these 
instances, the visible portion of the sea whip was mea- 
sured and recorded as a minimum measurement, but 
was not excluded from the analysis. 
Statistical analysis 
We compared the densities of invertebrates and physi- 
cal bottom features, such as hagfish burrows and trawl 
tracks, between LT and HT sites (Fig. 1) using analysis 
of variance (ANOVA). Trawling history (LT or HT) and 
block (northern or southern station pair) were treated as 
main effects with interaction, and transects were treated 
as replicate samples. Although our primary interest was 
the effect of trawling on seabed structure and structure- 
forming invertebrates, we included in our statistical 
analysis all fish and invertebrate groups that were 
encountered in at least 50% of the transects, according 
to the approach of McConnaughey et al. (2000). Density 
data were transformed by using Box-Cox transforma- 
tions with a displacement of 0.001 before ANOVA. The 
Box-Cox transformation employs a log-likelihood func- 
tion to find a power transformation that best normalizes 
the data (Sokal and Rolf, 1981; McConnaughey et al., 
2000). The residuals from each ANOVA were tested for 
normality with the Shapiro-Wilk goodness-of-fit test. 
Box-Cox transformations successfully normalized the 
density data for all of the taxa we compared statistically, 
with the exception of Dungeness crab (Cancer magister) 
and hagfish, as well as for the physical features we enu- 
merated (trawl tracks and hagfish burrows). Sea whip 
length data were not adequately normalized by trans- 
formation ( P< 0.01). For data that were not successfully 
normalized (P>0.05), nonparametric Wilcoxon rank sum 
tests were applied to the northern and southern station 
pairs separately by using JMP® 6 statistical software 
(SAS Institute, Inc., Cary, NO. To compare the inver- 
tebrate species diversity between LT and HT sites, we 
calculated the Shannon-Wiener diversity index (Pielou, 
1969) as well as species richness, defined as the number 
of species or taxonomic groups encountered at each site 
per unit of area sampled (Fig. 1). 
Results 
Hagfish burrows were very abundant and were the domi- 
nant structural feature of the sediment surface at all 
four sites (Fig. 3A). These burrows were mounds in the 
seafloor, with cone-shaped depressions in the center, 
ranging in diameter from approximately 10 to 60 cm. We 
