Ecological prediction using causal Bayesian networks: A case study of eutrophication management in the Neuse River estuary

We present a flexible method for policy-oriented ecological prediction based on a Bayesia n network. The graphical structure explicitly represents cause-and-effect assumptions between ecosystem variables that may be obscured under other approaches. These assumptions allow the complex causal chain linking management actions to ecological consequences to be factored into an articulated sequence of conditional relationships. Each of these relationships can then be quantified independently using an approach suitable for the type and scale of information available. Probabilistic functions describing the relationships allow key known or expected mechanisms to be represented without the full complexity, or information needs, of highly reductionist models. To demonstrate the application of the approach to a specific ecological problem, we present a detailed case study concerning eutrophication management in the Neuse River estuary, North Carolina. A Bayesian network was developed beginning with ecosystem attributes determined to be of direct interest to water quality stakeholders in the Neuse River watershed. These were then linked to anthropogenic nutrient inputs using causal scientific knowledge obtained at multiple scales. Relationships among variables were quantified using a variety of methods, including: process-based models statistically fit to long-term monitoring data, Bayesian hierarchical modeling of cross-system data gathered from the literature, multivariate regression modeling of mesocosm experiments, and probability judgments elicited from scientific experts. The integrated network was then used to generate predictions of the policy-relevant ecosystem variables under alternative nutrient management strategies. Predictions expressed as probability distributions give stakeholders and decision-makers a realistic appraisal of the chances of achieving desired outcomes. In general, the further down the causal chain a variable was, the greater the predictive uncertainty. This suggests that a compromise is necessary between policy relevance and predictive precision, and that, to select defensible environmental management strategies, Borsuk et al. – Ecological prediction and Bayesian networks, Page 3 of 3 public officials must adopt decision-making methods that deal explicitly with scientific uncertainty.