Lopez et al.: A model for estimating biomass of fish species associated with fish aggregating devices 
171 
Map of the 21 locations (•) where echosounder buoy acoustic data and catch information were collected 
around drifting fish aggregating devices (DFADs). Acoustic samples were taken at sunrise and were 
followed by regular fishing sets to gather information on the size and composition of catch. Fishing 
sets were conducted by a commercial Spanish purse seiner in the central and eastern Atlantic Ocean 
between 2009 and 2011. 
iV(gr) = l. n N(n,gr) (2) 
The estimated number of fish per group (Mgr]) was 
converted into biomass per group (B[gr], in t) by mul- 
tiplying the total amount of individuals by their corre- 
sponding mean weight (w, in kg) and dividing by 1000. 
1000 
where B(gr) = the biomass estimated per fish group (in 
t); 
Nigr) - the number of individuals per group; and 
w(gr ) = the average weight of an individual of a 
particular group (in kg) used to con- 
vert number of individuals in weight. 
Finally, the total uncorrected predicted tuna biomass 
(B u , in t) is the sum of the biomass estimated for the 
two tuna categories (corresponding to the sum of depth 
layers 3-10), whereas total biomass of nontuna species 
is the estimate obtained for that specific group (sum of 
layers 1-2). 
Parameter optimization Because vertical limits (depth 
boundaries) previously set may vary depending on the 
regional site, we used a set of 21 fishing sets to opti- 
mize our method and evaluate the depth limit selected 
for the particular case of the Atlantic Ocean. Addi- 
tionally, the same set of catches was used in a further 
stage to adjust and correct biomass estimates of the 
new method (see next section). 
Twenty-one acoustic samples collected from echo- 
sounder buoys between 2009 and 2011 in the central 
and eastern tropical Atlantic Ocean were used, along 
with their corresponding catch data, to make a first 
assessment of the performance of the proposed method 
(Fig. 3). Acoustic samples were collected at sunrise ( i.e. , 
the time of the day at which, according to the belief of 
fishermen, fish are supposed to be more concentrated 
under the DFAD) and were followed by regular fish- 
ing sets. We assumed purse seine catch to be a proxy 
for the total associated fish biomass at DFADs. Fishing 
operations were conducted by a Spanish purse seiner 
during conventional commercial fishing trips. Informa- 
tion on the species composition of the aggregation was 
then obtained in t and as commercial categories for 
each tuna species from the skipper’s logbook. 
In order to investigate the effect of the changes in 
the selection of the depth limit between small and 
large tunas for final biomass estimates, the echo-in- 
tegration procedure was conducted repeatedly by ap- 
plying all possible combinations to the depth limit be- 
tween small and large tunas within the entire depth 
range (i.e., changing the virtual limit from 80 m to 92 
m, 104 m, 115 m, 70 m, 59 m, 47 m, 36 m, and 25 
m), with the exception of the first 25 m, which were 
always considered inhabited mainly by nontuna spe- 
cies (Robert et al., 2013; Forget et al., 2015). For the 
