Evaluating the sensitivity of a trophic mass-balance model (Ecopath) to imprecise data inputs

Ecopath with Ecosim has become the most widely adopted modelling framework to facilitate a more holistic, ecosystem-based approach to fisheries management. I evaluated the sensitivity of Ecopath predictions to uncertainty in input data and determined the extent to which they were improved via the mass-balance constraint. I used nine published Ecopath models as operational models, added error to input variables, and then determined how well Ecopath could uncover the true model structure. I focused on Ecopath's ability to precisely estimate biomass and ecotrophic efficiency (the proportion of a group's total mortality explicitly represented in the model by predation and fishing). by Ecopath error predictions were approximately as precise as the input variables, except in cases of tightly linked cycles of food web connections when errors become greatly amplified. The process of model balancing had little effect on the magnitude of prediction errors. Moreover, accurate information about the source of imprecise data had little impact on prediction precision. Ecopath predictions were most sensitive to biomass and production rate input data and only occasionally sensitive to consumption rate and feeding habits data. Taken together, these results suggest that the precision of Ecopath-derived estimates of biomass and ecotrophic efficiency is generally comparable with the precision of the input data.