Improving the sampling protocol of electronic and human observations of the tropical tuna purse seine fishery discards

Observer programs have been implemented for many years in tuna purse seine fisheries to assess their impact on pelagic ecosystems by monitoring tuna discards and bycatch among which sensitive species such as sharks or rays. On board observers estimate discards using sampling and extrapolation methods when counting exhaustively is not possible. However, the flow of discards may be heterogeneous on the discard belt, and as a result, extrapolations may lead to over/underestimated estimations. Electronic monitoring system (EMS) on tuna fishing vessels has been tested as an alternative technology to complement and improve on board observer programs. EMS allows monitoring discards (of tuna and non-target species) at an acceptable species identification level and allows exhaustive counts on the discard belt. In this study, we used EMS “counts per minute” data from four French and one Italian purse seine vessels operating in Indian Ocean to evaluate total discards in numbers, as well as discards by species for each set. We analysed 48 fishing sets realised in 2017 and simulated different observer sampling strategies in order to optimise (i) the total sampling duration and (ii) the duration of sampling sequences. We finally propose an optimized sampling strategy, applicable to both electronic and human observations, for evaluating discards that reduces both sampling time and estimation bias.