176 
Fishery Bulletin 1 14(2) 
tion on the relative presence of both small and large 
tunas at DFADs. However, and likely due to the limited 
catch sample and the high numbers of small tunas oc- 
curring in those catches (i.e., -95%), the optimization 
process led us to keep a single group of tuna sizes for 
the whole vertical range. We expect that, by tuning and 
refining this method, by ocean or regional site and with 
greater amounts of data in the future, this approach 
will be able to remotely provide fishing crews and sci- 
entists with the potential proportion of different tuna 
sizes under a particular DFAD, with consequent fishing 
advantages and conservation applications. 
Model improvement 
Although FADs tend to standardize the vertical be- 
havior of fish (Matsumoto et al. 4 ; Taquet et al., 2007b; 
Leroy et al., 2009; Govinden et al. 5 ; Matsumoto et al., 
2012; Schaefer and Fuller, 2013), it seems likely that 
the main depths occupied by fish of different sizes and 
species may differ between oceans or regional sites. 
Because vertical behavior is likely affected by the 
oceanography of an area, future tagging experiments 
should cover unsampled locations and couple both si- 
multaneous in situ environmental data recorders (i.e., 
CTD profile recorders or acoustic surveys with scien- 
tific echosounders) and acoustic telemetry. High quality 
and high resolution data would contribute to a better 
understanding of the effect of biotic and abiotic factors 
on the vertical behavior of fish species when associ- 
ated with floating objects — research that is necessary 
to improve the interpretation of data from echosounder 
buoys. 
The present study employs the TS values obtained by 
Moreno et al. (2007), which fit reasonably well with the 
preliminary vertical depth layers defined in this work 
and with the results of other studies (Josse and Ber- 
trand, 2000; Josse et al., 2000; Doray et al., 2006; Doray 
et al., 2007). As is widely recognized, knowing the TS 
value of a species is of primary importance to properly 
transform acoustic backscatter into species biomass. 
However, TS of many FAD-associated species remains 
elusive. Further ex situ or in situ investigations cou- 
pling acoustics with other supplementary technologies, 
such as video-cameras or ROVS (i.e., remotely operated 
vehicles) would permit obtaining precise TS values by 
species and thus improve the accuracy of estimates at 
species levels. Nonetheless, species-specific biomass 
conversion seems to be difficult to achieve in the cur- 
rent scenario. As Handegard et al. (2012) stated, one of 
the major challenges associated with acoustic measure- 
ments on autonomous platforms is the lack of biologi- 
cal sampling to verify the taxonomic composition. One 
possible solution to solve this situation is to explore 
catch variability and obtain spatiotemporal patterns of 
species composition at DFADs. Results could be used in 
future experiments to improve the performance of the 
method by postprocessing actual biomass estimates in 
cases where no biological sampling is available. 
Although the echosounder buoys used in this study 
were developed for fishing purposes, there is still much 
room to improve the technical and technological fea- 
tures of current echosounder buoys. Ideally, some tech- 
nical issues such as the lack of regular calibrations of 
the transducers or the characterization of their vertical 
and angular detection ranges should be solved. These 
are complicated issues due to the large number of echo- 
sounders (thousands of each brand) that are deployed 
with DFADs each year globally. A compromise solution 
could be the calibration of a limited number of trans- 
ducers of each brand along an extended period of time, 
long enough to model a general calibration factor as a 
function of time. If consistent, the obtained brand-spe- 
cific function could be then applied to all transducers, 
replacing the need to calibrate each transducer for ev- 
ery period of time. These calibrations could also include 
detectability functions at different depth ranges and 
angles in order to characterize the vertical and angu- 
lar detection ranges of the transducer as a function of 
the TS of the targets. Then, the obtained detectability 
functions could be applied by using distance-sampling 
techniques (Buckland et al., 1993). Similarly, the range 
of the echosounder system, operation frequency, split 
beam technology, etc. should also be re-examined, opti- 
mized, installed, and adapted to the presence of differ- 
ent tropical tuna and accompanying nontuna species. 
Therefore, continuous improvements would be possible 
in the successive releases of new generations of trans- 
ducers. For example, the recent inclusion of multifre- 
quency transducers in two echosounder buoy brands 
could enable improved species discrimination in the 
near future, although those transducers would also re- 
quire improved algorithms for analysis. 
Due to confidentiality and sensitive issues, the flow 
of echosounder buoy data from vessel owners to re- 
search institutes is still rare. A data exchange system 
that preserves the privacy and ensures future bilat- 
eral advantages could help in effective industry and 
science collaboration. These kinds of initiatives should 
be endorsed by international bodies (e.g., tuna regional 
fisheries management organizations) to guarantee the 
viability of a project for both industry and science and 
to help promote such studies with an aim toward main- 
taining the long-term sustainability of fisheries. 
Acknowledgments 
The authors sincerely thank all the fishing masters 
from the Spanish tuna purse-seiner fleet who kindly 
agreed to answer our questions. We would like to thank 
the Spanish ship-owners and the associations that rep- 
resent them (Asociacion Nacional de Armadores de 
Buques Atuneros Congeladores [ANABAC] and Orga- 
nizacion de Productores Asociados de Grandes Atune- 
ros Congeladores [OPAGAC]). Of course we would like 
to thank the buoy manufacturers that provide us with 
technical data on their echosounder buoys. We sin- 
cerely thank G. P. Scott for revising the English and 
for useful comments that considerably improved this 
