Uncertainty and the Use of Network Analysis for Ecosystem-Based Fishery Management

The fisheries management community is moving away from single species management and towards an ecosystem-based approach. Inherent to this effort is an understanding of the interactions between harvested species, their prey, predators, and competitors. One approach increasingly being employed is ecological network analysis (ENA), a modeling technique most often used to examine food webs. This article has three objectives: (1) to call attention to the potential of ENA for ecosystem-based fisheries management, (2) to identify uncertainties associated with ENA, and (3) to provide recommendations for addressing these uncertainties. ENA provides a method for quantifying direct and indirect trophic interactions, for comparing food web properties among different systems and/or times, and for incorporating fishery harvest into the analysis. However four major sources of uncertainty exist: natural variability of input parameters, data collection methods, model construction, and fundamental assumptions of the algorithms. Few ENA studies address uncertainty in their models and even fewer validate model output. A priori predictions of model output and sensitivity analysis could be used to better understand the effect of variability in model input. Model construction could be improved by incorporating multivariate techniques, and concerns of how well a model depicts the real-world system could be addressed by validating model output with independent techniques. No single index from ENA gives a comprehensive picture of trophic conditions and uncertainties may cloud prediction. However when model output is considered as a group, ENA provides insight into the structure and functioning of the entire ecosystem and hence is useful for ecosystem-based fisheries management.