Robert P. Wildermuth1,2, Gavin Fay1, Sarah Gaichas3 and Geret DePiper3
1. Department of Fisheries Oceanography, School for Marine Science and Technology, University of Massachusetts Dartmouth, New Bedford, MA, USA.
3. NOAA National Marine Fisheries Service, Northeast Fisheries Science Center, Woods Hole, MA, USA.
Integrated ecosystem assessment requires decision tools that can use multiple datasets, account for uncertainties, and assess tradeoffs among ecological, social, and economic objectives. Several environmental stressors and human activities affect fisheries on Georges Bank, USA. Tradeoffs among multiple ocean uses and the multitude of benefits provided to the Georges Bank socio-ecological system are difficult to assess with extant quantitative models due to imbalanced data availability across system components. We developed a Bayesian network model for the Georges Bank social-ecological system that integrates monitoring data and expert knowledge to estimate direct and indirect effects of multiple human uses of living natural resources for a set of management objectives. We used maximum likelihood methods and a survey of social science experts to fit our model to a 58-year time series for Georges Bank and evaluated the modelís predictive ability. We also analyzed the influence multiple scenarios had on the perceived ability to meet 12 management objectives. Altogether, the model correctly predicted component states greater than 70% of the time, though it poorly anticipated component state transitions. Our evaluation revealed potential leading indicators for habitats and managed functional groups, as well as unexpected outcomes for seafloor and demersal habitat. We demonstrate 1) how to integrate expert knowledge in the development and evaluation of Bayesian network models, and 2) how these methods can provide strategic management advice for complex marine social-ecological systems, even when few data exist for key components.