226 
Fishery Bulletin 119(4) 
attraction and avoidance functions that in future itera- 
tions of this research will be useful for conducting vehicle 
calibration studies and for modifying fish abundance and 
density estimates. 
The benthic and pelagic pursuers guilds had strong 
attraction to the survey vehicles. Both guilds included sev- 
eral managed species that are important components of 
fisheries that operate in the GOM, including scamp, red 
grouper (Epinephelus morio), and greater amberjack. In 
addition to the obvious distinction of habitat preference, 
these guilds had distinct differences in ability to match 
vehicle speed while in pursuit. Pelagic pursuers concen- 
trated behind vehicles and could easily maintain pace 
with them; however, their response was not persistent over 
repeated transects (Fig. 5). The strong attraction signal, 
coupled with the pelagic pursuers ability to match vehicle 
speed, indicates that the same fish could be observed with 
vehicles along the transect line, resulting in overestima- 
tion of fish abundance or density. Unfortunately, given the 
relatively short transect length in our surveys, we could 
not determine the distance over which pelagic pursuers 
continued to follow a vehicle. 
Conversely, benthic pursuers gave short chase following 
vehicle passage and could not keep up with vehicles; how- 
ever, they repeated the behavior in subsequent vehicle tran- 
sects (Fig. 6). This finding indicates that new individuals 
from the benthic pursuers group are likely being observed 
with the vehicles down the transect line. Thus, density esti- 
mates for species of this guild likely will not require adjust- 
ment because the observed fish did not aggregate around 
the vehicles and, as a result, were not drawn down a tran- 
sect line. Of higher concern for observing benthic pursuers 
guild fish like groupers is their coloration and mottling pat- 
terns (Burge et al., 2012; Watson et al., 2014) and their ten- 
dency to hide underneath structure, both of which makes 
them difficult to observe (Camp et al., 2013). Therefore, 
understanding detection probability and sighting functions 
may be more critical for estimating abundance or density 
for species in the benthic pursuers guild. In our experi- 
ment, the following behavior seems to be most strongly 
associated with piscivorous fish, perhaps as a stalking or 
investigatory type of behavior. Strong attraction to ROVs 
has also been observed for yellowtail rockfish (Sebastes fla- 
vidus) (Stoner et al., 2008) and Pacific hake (Merluccius 
productus) (Adams et al., 1995); however, neither study 
quantified the response. In general, attraction responses 
need to be well understood because they concentrate fish, 
could result in overestimation of density, and are thus not 
conservative. 
Gear avoidance is one of the more common behaviors 
noted in studies (Adams et al., 1995; Koslow et al., 1995; 
Uiblein et al., 2003; Lorance and Trenkel, 2006) and was 
most clearly observed for the lateral and down movers 
guilds. Lateral movers guild fish, which include snapper 
and grunts, exhibited significant decreases in abundance 
with decreasing vehicle range, indicating a flight response 
to approaching vehicles. Avoidance behaviors have been 
linked to the looming effect in which fast, direct approaches 
increase the likelihood and strength of the flight response 
(Frid and Dill, 2002). An analysis of fine-scale data col- 
lected from frame-by-frame analysis of a subset of the 
UHSI videos revealed that vermilion snapper scaled 
their response from slow-swimming behavior and loosely 
aggregated schools when the vehicle was distant to tightly 
aggregated schools and fast-swimming behavior when the 
TV came into the sampling volume (Somerton et al., 2017). 
Both the ROV and TV displaced lateral movers guild fish 
from the MOUSS platform sampling volume prior to the 
vehicle’s arrival; however, there was not enough AUV data 
to analyze. The persistence of the flight response was vehi- 
cle dependent; for instance, abundance quickly returned 
to pre-disturbance levels following TV transit; whereas, 
abundance remained depressed following ROV transit. The 
difference in the persistence of the effect could be related 
to the extended length of time the ROV tether remained 
in the vicinity of the MOUSS platform. Calibration experi- 
ments will be necessary for vehicle-based surveys that tar- 
get fish species in this guild and that exhibit strong flight 
response as there is high likelihood that abundance and 
density will be underestimated. 
Another clear case of vehicle avoidance was associated 
with the down movers guild. The flight response to the 
approaching vehicle was consistent and strongly oriented 
downward into protective habitat, but fish did not depart 
the sampling volume (i.e., present but no longer detected). 
Most of the species in the down movers guild in this anal- 
ysis are in the family Pomacentridae (e.g., Chromis spp.), 
are generally small in body size, and are found in close 
proximity to reef habitat. The distinct downward response 
to the vehicle was also observed when predators, such as 
great barracuda swam through the sampling volume, and 
thus we presume this behavior to be related to predator 
avoidance and response to novel threats. It could be that 
the body size of the species in this group precludes them 
from outswimming a predator in open water and that a 
strategy to stay near protective habitat into which they 
can flee is advantageous. In addition, the response appears 
to be weakest in relation to the AUV and strongest for the 
TV and ROV. As with the response of the lateral movers, 
the looming effect is strongly associated with the speed 
of the vehicle (Frid and Dill, 2002) and might explain 
the observed differences in fish responses between vehi- 
cles. The AUV travels slower and at higher altitudes than 
the other vehicles and therefore has a decreased looming 
effect, which perhaps tempers the flight response into pro- 
tective habitat (i.e., perceived threat is dampened). As with 
the lateral movers guild, gear calibrations are required for 
species in this group to compensate for changes in abun- 
dance due to the flight response. 
The local reactive guild is composed largely of species 
of ornamental reef fish, such as butterflyfish (Chaetodon 
spp.), filefish (Wonacanthus spp.), and damselfish (Ste- 
gastes spp.). For this guild, abundance increased as a 
function of vehicle range, and fish of this guild exhibited 
little attraction or avoidance to the vehicles and, as with 
fish of the down movers guild, did not leave the sampling 
volume. Review of video also revealed increased activity 
(i.e., reactive behavior) following vehicle passage. Further, 
