32 
Fishery Bulletin 108(1 ) 
We used trawl logbook data from the state of Oregon 
as one indicator of the history of trawling at the four 
study sites. Data for vessels that landed shrimp in 
other states were considered minimal and not included 
here because of a lack of availability of Washington 
and California ocean shrimp logbook data after 1992 
(Hannah, 1999). The logbook information provides 
only haul start locations and not actual trawl paths; 
therefore it must be regarded as a crude index of the 
level of trawling at these sites. However, the pattern of 
the data that is available indicates that all four study 
sites have been predominantly impacted by ocean 
shrimp trawls. Ocean shrimp trawl start locations 
(starts) within these four sites from 1988 through 
2007 averaged between 0.65 and 8.25 starts/yr (means 
for 5-year periods are shown in Fig. 2A), whereas 
groundfish trawl starts/yr averaged from 0.05 to 0.35 
starts/yr. The dominance of ocean shrimp trawling 
at these sites is also understated by these averages 
because the available shrimp trawl logbook data are a 
subsample of the total ocean shrimp trawl effort (26- 
75% of total effort included; Hannah, 1995) whereas 
the groundfish trawl data represent a more complete 
census. The logbook data also show a large difference 
in ocean shrimp trawling history between each of the 
site pairs with similar latitude (between 1A and IB, 
and between 2A and 2B in Fig. 2A). For the northern 
pair, the easternmost site (IB) has received on aver- 
age, about seven times more ocean shrimp hauls than 
the western site (1A). The easternmost southern site 
(2B) has received an average of about six times more 
ocean shrimp hauls than the more western site (2A). 
Hereafter the two eastern sites will be referred to as 
heavily trawled (HT) and the western sites as lightly 
trawled (LT). 
To conduct seafloor habitat surveys of the four sites, 
we chartered the 21-m shrimp fishing vessel Miss 
Yvonne, out of Newport, OR. Video surveys were con- 
ducted with a Phantom HD2+2 (Deep Ocean Engi- 
neering, San Leandro, CA) remotely operated vehicle 
(ROV) during 20-23 June 2007. Using this system, we 
gathered standard-resolution video and GPS location 
data along each of 21 transects (Fig. 1), which were 
viewed and recorded in real time. The transects were 
spaced systematically across each site; however, actual 
transect paths were dependent on the prevailing direc- 
tion of vessel drift. Thus, they varied in orientation and 
in some instances were divided into segments because 
of unusable portions (Fig. 1). Lighting was provided by 
two 200W lights (Nuytco Research Limited Newtlite, 
North Vancouver, BC). This system is also equipped 
with two parallel forward-facing lasers (15 mW, red), 
situated 10 cm apart and aligned with the central axis 
of the primary camera (forward facing oblique), that can 
be used for scaling the field of view and objects viewed 
by the camera. 
All four sites at Nehalem Bank had similar depth 
ranges (Table 1). The areas and linear distances sur- 
veyed were also similar, with the exception of site 2B, 
where an equipment malfunction resulted in a lack of 
□ 1988-92 
■ 1993-97 
1 A IB 2A 2B 
Sample site 
Figure 2 
(A) Mean ocean shrimp ( Pandalus jordani) trawl 
starts per year (±1 standard error [SE J ) for four 5- 
year periods from subsampled fishery logbook data, 
and (B) the density of trawl tracks (tracks/ha, ±1 
SE) observed with a remotely operated vehicle for 
four sites with different histories of ocean shrimp 
trawling at Nehalem Bank, Oregon (see Fig. 1). 
ROV position data for a portion of two transects, reduc- 
ing the linear distance surveyed to 7.11 km in compari- 
son to about 9.30 km for each of the other sites. 
Video analysis 
Processing of video data began with an initial review of 
the tapes to identify usable segments for quantitative 
analysis. Only short segments of video proved unusable, 
because of poor visibility, or because the ROV was either 
too far above the bottom, off the intended transect, or 
briefly pulled sideways by the research vessel. This ini- 
tial review identified start and stop times of each usable 
segment for subsequent quantitative analysis (Table 1). 
Transects for which we report data are composites of 
these large segments of usable video data. 
Organisms, and other features, including trawl tracks 
and biogenic features such as hagfish (Eptatretus spp.) 
burrows, were enumerated from the video only if they 
were in view at the vertical center of a standard video 
screen (50% line), where the lasers were also aligned. 
This approach avoided an unrealistic extension of the 
top of the camera’s estimated field of view beyond the 
