Campbell et al.: Attraction and avoidance behaviors of fish in response to proximity of underwater vehicles 221 
Thus, developing species-specific models was not possible 
for most species observed; however, general attraction and 
avoidance response behaviors were easily characterized. 
As a result, species were grouped into 5 behavioral guilds, 
and abundance data from the species belonging to those 
groups were pooled for analysis (Table 1). Behavioral 
guilds identified and analyzed included pelagic pursuers, 
benthic pursuers, down movers, lateral movers, and local 
reactive. Group composition is shown in Table 1 but in 
brief was as follows: the pelagic pursuers were composed of 
jacks (Carangidae) and mackerels (Scombridae); the ben- 
thic pursuers were composed of groupers (Serranidae) and 
wrasses (Labridae); lateral movers were composed of snap- 
pers (Lutjanidae) and porgies (Sparidae); down movers 
were composed of species of damselfish (Pomacentridae) 
and jawfish (Opistognathidae); and the local reactive guild 
was composed of species of filefish (Monacanthidae), but- 
terflyfish (Chaetodontidae), and sand perch (Diplectrum 
spp.). Fifteen models were produced for each unique com- 
bination of the 3 vehicles and 5 behavioral guilds. 
Changes in fish counts during the MOUSS platform 
acclimation time were evaluated by using generalized 
additive models (GAMs) with the GAM predict func- 
tion of the package mgcv (vers. 3.6.1; Wood, 2011) in R 
(vers. 4.0.3; R Core Team, 2020). Fish acclimation mod- 
els included the following variables: time (seconds) and 
habitat complexity (qualitatively rated 1-5 from low 
to high). Changes in relative abundance due to vehicle 
passage were also analyzed by using GAMs. Because all 
vehicles did not necessarily transit through equivalent 
transects (the TV and AUV had no shared transects), and 
therefore did not pass through equivalent underlying fish 
densities, changes were analyzed in separate models for 
each vehicle type (i.e., we did not create a single model 
with a vehicle variable). Additionally, the habitat complex- 
ity variable allowed us to model site-specific differences 
in abundance that could potentially mask responses to 
a vehicle. For example, without a variable that explains 
site-specific differences, high-abundance sites could 
potentially have disproportionate effects on outcomes. 
Models used to examine fish responses to vehicle pas- 
sage included the following variables: RVA (low, middle, 
and high), transect number (integer), range (meters), and 
ranked habitat complexity (qualitatively rated 1-5 from 
low to high). Deployment of the entire transect course 
took significant effort. Thus, multiple transects were sur- 
veyed over the course of a day, and that daily number of 
transects was tracked as a sequential value called tran- 
sect number. Transect number therefore is used to compile 
the total number of vehicle exposures that occurred on the 
transect course over a single deployment of the 3 MOUSS 
platforms on the transect course. 
Table 1 
Behavioral guild assignments based on movements of fish taxa in response to 
presence of mobile survey vehicles deployed in the Florida Middle Grounds during 
August 2014 and July and August 2015. 
Pelagic pursuers 
Seriola dumerili 
Sphyraena barracuda 
Seriola rivoliana 
Carangidae 
Seriola sp. 
Caranx lugubris 
Scombridae 
Lateral movers 
Rhomboplites aurorubens 
Lutjanus griseus 
Lutjanus campechanus 
Calamus calamus 
Calamus nodosus 
Calamus penna 
Calamus proridens 
Calamus sp. 
Haemulon plumierii 
Haemulon sp. 
Pagrus pagrus 
Clepticus parrae 
Benthic pursuers 
Mycteroperca venenosa 
Mycteroperca phenax 
Mycteroperca sp. 
Epinephelus morio 
Serranidae 
Halichoeres bathyphilus 
Halichoeres bivittatus 
Halichoeres sp. 
Local reactive 
Monacanthus ciliatus 
Stephanolepis hispidus 
Monacanthus sp. 
Monacanthus tuckeri 
Stegastes leucostictus 
Stegastes partitus 
Stegastes sp. 
Stegastes variabilis 
Chaetodon ocellatus 
Chaetodon sedentarius 
Chaetodon sp. 
Diplectrum formosum 
Decodon puellaris 
Down movers 
Pomacentridae 
Pomacentrus sp. 
Chromis insolata 
Chromis enchrysura 
Chromis sp. 
Opistognathus aurifrons 
Opistognathus sp. 
