Network approaches for formalizing conceptual models in ecosystem-based management

Abstract

Qualitative Network Models (QNMs), Fuzzy Cognitive Maps (FCMs), and Bayesian Belief Networks (BBNs) have been proposed as methods to formalize conceptual models of social-ecological systems and project system responses to management interventions or environmental change. To explore how these different methods might influence conclusions about system dynamics, we assembled conceptual models representing three different coastal systems, adapted them to the network approaches, and evaluated outcomes under scenarios representing increased fishing effort and environmental warming. The sign of projected change was the same across the three network models for 31-60% of system variables on average. Pairwise agreement between network models was higher, ranging from 33 to 92%; average levels of similarity were comparable between network pairs. Agreement measures based on both the sign and strength of change were substantially worse for all model comparisons. These general patterns were similar across systems and scenarios. Different outcomes between models led to different inferences regarding trade-offs under the scenarios. We recommend deployment of all three methods, when feasible, to better characterize structural uncertainty and leverage insights gained under one framework to inform the others. Improvements in precision will require model refinement through data integration and model validation.