Résumé
The multi-species nature of tropical tuna surface fisheries gives rise to a series of difficulties whenestimating the catch by species and catch at size statistics. The T3 processing was built about 30 yearsago in order to correct biases of the logbook data on species composition and to provide more accuratecatch estimates per species for the European purse seine fleet. However, the evolution of fishingpractices and the extension of the fishing grounds have challenged the T3 methodology on some partsof it processing. The aim of this paper is first to give the key elements to evaluate the potential biasesthat could occurs in the catch assessments of tropical tunas of purse seiners and second to exploresome ways in order to increase accuracy of catch estimates based on the T3 processing for the future.By comparing catch weights obtain from T3 processing output and from operational landing reportsproduced by fishing companies, we found a potential overestimation of catch of less dominant species,which lead to an underestimation of dominant species. This bias should be mainly due to the too largespatio-temporal sampling strata used to correct the catch by specific capture reported in logbooks aswell as, for a minor part, could be a consequence of the change over time in length-weightrelationships used in the T3 processing. We also discussed on the limitations and assumptions of theT3 processing in relation to the data quality and the reliability of the Operational landing reports.