Bacheler et al.: Influence of soak time and fish accumulation on the catches of reef fishes 
227 
Table 4 
Percent change in estimates of overall catch per trap for 8 reef fish spe- 
cies when soak time doubled (from 60 to 120 min) on the basis of data 
from 2 sampling programs in the southeastern U.S. Atlantic: the Marine 
Resources Monitoring, Assessment, and Prediction Program (1990-2011) 
and the Southeast Fishery-independent Survey (2010-11). Catch was es- 
timated with a delta-generalized-additive model where mean values of 
all other predictor variables were used. Species are arranged from high- 
est to lowest percent change in catch per trap. 
Species 
Estimated catch 
at 60 min 
Estimated catch 
at 120 min 
Percent 
change 
Red Porgy 
1.25 
2.57 
106 
Gray Triggerfish 
0.80 
1.56 
95 
Black Sea Bass 
6.45 
7.71 
20 
Bank Sea Bass 
3.51 
3.83 
9 
Vermilion Snapper 
2.79 
3.02 
8 
Sand Perch 
1.67 
1.24 
-26 
Tomtate 
6.78 
4.58 
-32 
Stenotomus spp. 
0.99 
0.29 
-71 
times examined, and, for others, be- 
cause of fish accumulation in the trap. 
A number of methods have been de- 
veloped to determine if gear saturation 
is occurring. Addison and Bell (1997) 
used a simulation approach to show that 
the relationship between lobster catch 
and abundance was asymptotic, a prob- 
lematic result because models would 
predict an even spatial distribution of 
lobster catches across a study area de- 
spite a true underlying aggregated dis- 
tribution. Some researchers have docu- 
mented gear saturation by the fact that 
the cumulative catch of individuals in 
traps that are periodically emptied is of- 
ten much higher than the catch in traps 
that were hauled and redeployed with- 
out being emptied (Miller, 1979; Robert- 
son, 1989). Alternatively, Li et al. (2011) 
developed a delta-GAM to quantify the 
relationship between catch of Walleye 
(Sander vitreus) and Yellow Perch and 
soak time or fish accumulation at average values of all 
other covariates in the model. We used the Li et al. 
(2011) modeling approach to show that chevron traps 
became saturated for all 8 reef fish species examined 
across a range of values for soak time or fish accumula- 
tion. There are 3 major benefits of this approach. First, 
it is possible to test for trap saturation through the use 
of long-term survey data, as long as there has been suf- 
ficient variation in soak time. Second, the relationship 
between catch and soak time or fish accumulation can 
be quantified after accounting for variation due to the 
other predictor variables in the model. Last, zero-in- 
flation, the situation where a large number of zero ob- 
servations in a data set cannot be properly accounted 
for with traditional statistical distributions, can 
be properly accounted for through the use of a delta 
model. 
Our study was improved by the inclusion of the fish- 
acc variable in the models. If the rate at which a spe- 
cies entered the trap was unaffected by the number of 
individuals (of all species) already caught in a trap, 
then one would expect a positive, linear relationship 
between the catch of a species and the fishacc variable 
(Li et al., 2011). We showed that catches of Red Porgy, 
Bank Sea Bass, and Gray Triggerfish plateaued once 
a moderate number of total individuals were already 
caught in a trap, indicating that these species are 
more sensitive to species interactions and, therefore, 
much less likely to enter a trap once it began filling up. 
These results are consistent with previous work that 
has shown that behavioral interactions in and around 
traps can strongly influence the catch of target species 
(Addison and Bell, 1997; Jury et al., 2001; authors, per- 
sonal observ. ). In contrast and, perhaps, more surpris- 
ingly, Black Sea Bass, Tomtate, Vermilion Snapper, and 
Sand Perch continued to enter a trap at about the same 
rate no matter how many total individuals of all spe- 
cies were caught in it. A primary benefit of inclusion of 
a predictor variable for fish accumulation in our model 
was that it allowed us to distinguish between species 
that saturated the gear because of fish accumulating 
in a trap from the species that appeared to saturate 
the gear because of the amount of time a trap soaked. 
In addition, the inclusion of the fishacc variable stan- 
dardized the catch of each of the 8 reef fish species 
to a common total catch of all species in the trap. In 
other words, we were able to remove variations in the 
catch of each reef fish species that were attributable 
to species-specific responses to fish accumulation. The 
inclusion of a variable for fish accumulation in a stan- 
dardization model is one straightforward approach that 
can be used to account for some species interactions. 
Our results indicate that catch per trap, not catch 
per trap minute, should be the response variable used 
in future standardization models for all the species we 
examined. If catch is invariant to soak time and catch 
per unit of effort (CPUE) is used as the response in 
a catch standardization model, then CPUE on average 
would be lower in traps with longer soak times than 
in traps with shorter soak times. Soak-time-dependent 
CPUE could become a serious problem if soak time for 
traps was longer in some years than in others because 
real changes in the relative abundance of a species 
would be confounded with the effects on CPUE due to 
changes in soak time. Instead, we recommend the use 
of a model-based approach with catch as the response 
variable and soak time and fish accumulation as pre- 
dictor variables to properly account for any variation in 
catch due to these 2 factors. 
Whether catch data from trap surveys can be used 
to index reef fish abundance ultimately depends on the 
underlying mechanisms responsible for the observed 
