“…This issue has been historically addressed in three ways: (1) by the so-called "output-based" methods (Marchal, 2008), consisting of a posteriori quantitative analyses of catch or landings composition, with or without effort information (gear, season, location) to assign métier. Various multivariate procedures have been applied for this purpose: principal component analysis (PCA; Castro et al, 2010;Jabeur et al, 2000;Laurec et al, 1991), multiple correspondence analysis (Pelletier and Ferraris, 2000), and cluster analysis (Lewy and Vinther, 1994); (2) by the so-called "input-based" (Marchal, 2008) methods, based on a priori qualitative knowledge of the fisheries, mainly collected during face-to-face interviews, so that the allocation of each fishing trip to métier relies on a process of trial and error, by deriving discriminating thresholds based either on landings (weight or value), or mesh size (Biseau, 1998;Tétard et al, 1995;Ulrich and Andersen, 2004); or (3) by a combination of the input and output-based methods (Bastardie et al, 2010;Ulrich and Andersen, 2004). However, both types of method (as well as their combinations) have some limitations: they depend on available data, and landings data may be unreliable and may not contain enough information to infer fishing intention (Chang, 2011;Marchal, 2008).…”