Campbell et al.: Attraction and avoidance behaviors of fish in response to proximity of underwater vehicles 219 
Surface buoy 
Anchor 
MOUSS platform 
61m 
122m 
244m 
Figure 3 
General schematic of 3 ground-tended Modular Optical 
Underwater Sampling System platforms (rectangles) set 
61 m apart along a groundline. Anchors (triangles) and 
surface buoys (circles) were attached 61 m from the plat- 
forms at each end of the line. Total length between anchors 
was 244 m, and the total distance between end cameras 
was 122 m. 
were 18 cm above the seafloor. Cameras have a 70° field 
of view (FOV) and use ambient light, and camera settings 
were optimized for expected light conditions in clear, 
shallow water (20—40 m). Still images were acquired at a 
rate of 5 frames per second and were later processed to 
make video files in MPEG format. Geographic position of 
each platform was determined by using repeated acoustic 
surveys that triangulated the LBL acoustic beacons, and 
those positions were used to aid navigation of each vehi- 
cle in front of the MOUSS platforms. 
To ensure that the orientation of each of the 3 MOUSS 
platforms was consistent, they were deployed in a fashion 
similar to that of demersal longline traps, such as those 
used in the sablefish (Anoplopoma fimbria) fishery in 
U.S. waters (Afanasyev et al., 2014) (Fig. 3). Platforms 
were attached to a 9.5-mm bottom line that was continu- 
ously spooling off a winch while the ship maintained a 
constant heading during setting. The entire set was 
anchored on each end by steel weights (22 kg) located 
61 m from the 2 outermost platforms. Two quick links 
were spliced into the longline at predetermined points 
61 m apart, onto which the platforms were attached with 
high-strength snap shackles that were connected to plat- 
form harnesses. Anchor to anchor, the transect distance 
was 244 m, and the distance between end point cameras 
was 122 m (Fig. 3). This method of deployment allowed 
the main line to act as a yoke to keep the MOUSS plat- 
forms, and therefore camera orientation, roughly perpen- 
dicular to the ship’s bearing. Cameras were normally 
deployed in ~30 min, and then the LBL beacons were sur- 
veyed to triangulate MOUSS platform position over a 
period that was typically 2-3 h. The deployment and tri- 
angulation process provided time for fish to become accli- 
mated to the presence of the MOUSS platforms prior to 
the deployment of the survey vehicles. 
Vehicles and transects 
Survey vehicles evaluated in the experiment included a 
Phantom HD2+2' ROV (Deep Ocean Engineering, San 
Jose, CA), the Camera-based Assessment Survey System 
TV (Lembke et al., 2017), and the SeaBED AUV (Singh 
et al., 2004). For simplification we refer to the vehicles we 
used as ROV, TV, and AUV. Each vehicle has different 
deployment and operation properties that limited our abil- 
ity to stage all vehicles simultaneously on the survey ship 
and to deploy all vehicles in a single day. Therefore, the 
survey was divided into 2 segments that were conducted 
back to back. As a result, general deployment areas could 
be surveyed with each vehicle, but exact locations were 
not replicated across survey legs, resulting in each vehicle 
operating in somewhat different habitats (i.e., depth, sub- 
strate type, and fish community and abundance varied by 
site). Thus, models used to evaluate fish responses were 
somewhat vehicle specific. 
The ROV (140 x 69 x 66 cm, in length, width, and height, 
respectively) was constantly tethered to the ship to pro- 
vide power for vehicle operations and was remotely piloted 
from the ship by using a combination of 2 live-feed cam- 
eras (front and rear), sector scanning sonar, and a com- 
pass. During ROV deployments, the ship was anchored 
and, as a result, no tether management system was 
required and the tether was sufficiently long (338 m) to 
navigate from anchor to anchor at all operational depths 
during the experiment. Following ROV launch, the vehi- 
cle was driven on the surface to the closest down-current 
buoy, and then the ROV descended to the anchor. Upon 
reaching the seafloor, the pilot oriented the ROV on the 
desired transect heading and parked the ROV on the 
seafloor for 15 min prior to beginning to survey the first 
transect. Transects were run parallel to the orientation of 
the 3 MOUSS platforms and camera FOV, from anchor to 
anchor in both directions, and with 15-min resting periods 
between each transect. Depending on sea state and time 
of day, 2-6 transects were surveyed at each site prior to 
recovery of the vehicle. 
The TV (179 x 127 x 87 cm, in length, width, height, 
respectively) was towed behind the ship by using a winch 
and conducting wire that also powered the vehicle and 
provided real-time seafloor imagery to assist the pilot in 
navigation. Vehicle operations were conducted by a pilot 
who communicated with the ship’s officers on the bridge 
about preferred heading and speed, and with deck per- 
sonnel regarding winch payout and vehicle altitude. 
' Mention of trade names or commercial companies is for identi- 
fication purposes only and does not imply endorsement by the 
National Marine Fisheries Service, NOAA. 
