Influencia de las interacciones con el factor Año en los índices anuales de abundancia obtenidos por modelos lineales generales utilizando datos de Captura por Unidad de Esfuerzo

Abstract

The problem in the definition of a relative annual abundance index using a General Linear Model is posed. The possible factors generating changes in the catch per unit effort (CPUE) are discussed and the parameterization of the model is analyzed, in particular the Sigma-restrictive parameterization. This report considers the problem caused by the presence of interactions of first order that include the factor Year in the model. It has been mathematically demonstrated that in the presence of these interactions, the index series is not proportional to the abundance annual mean of the resource. An approach of practical order to evaluate the possible effect of the interactions that include the factor Year in the index is considered using the percentage of explained variance of the model with and without these interactions. Some methodological rules are also defined to attenuate several problems generated by the mentioned interactions: (1) evaluating if type of fleet, included as categories of the factor Ship, has experienced modifications throughout the period of study that could have generated changes in the efficiency of fishing units; (2) allowing for the possibility of working with indexes pondered by Area; (3) using the most convenient categorization for the factor Month (month, bimonthly, quarterly, fishery season, etc.), considering the seasonal distribution of the resource and the activity of the fleet, as well as the limitations generated by the presence of empty cells in the design. Two application examples are presented