Modelling the skipjack tuna dynamics in the Indian Ocean with APECOSM-E – Part 2: Parameter estimation and sensitivity analysis

Abstract

This paper presents the parameter estimation and sensitivity analysis of the APECOSM-E model, which describes the basin scale 3D size-structured population dynamics of the skipjack tuna population in the Indian Ocean under the joint effect of environmental variability and fisheries exploitation. The model is presented in detail in the companion paper (Dueri et al., 2012). A common methodology based on the evaluation of a cost function that combines the negative log-likelihoods of commercial catches and size frequencies is used for both tasks. A Bayesian term representing the a priori probabilities about the model parameters is added to the cost function used for parameter estimation. The partial derivatives of the cost function with respect to the parameters are obtained by deriving the tangent linear code of the model by automatic differentiation of the direct code. A minimization algorithm is used to estimate the parameters related to fisheries and maximise the fitness of the model to the available observations. In a second step, we evaluate the local sensitivity of the non-estimated parameters and identify the model parameters that have an important effect on the output of the model and that would deserve better measurements in order to reduce the level of uncertainty in the model outputs. The comparison between the optimized simulation and the observations allows identifying the model's strengths and limitations, in the perspective of using the model to test scenarios concerning the resilience of the population and the sustainability of its exploitation in the Indian Ocean.