Integrating modeling, monitoring, and management to reduce critical uncertainties in water resource decision making.

Stream ecosystems provide multiple, valued services to society, including water supply, waste assimilation, recreation, and habitat for diverse and productive biological communities. Managers striving to sustain these services in the face of changing climate, land uses, and water demands need tools to assess the potential effectiveness of alternative management actions, and often, the resulting tradeoffs between competing objectives. Integrating predictive modeling with monitoring data in an adaptive management framework provides a process by which managers can reduce model uncertainties and thus improve the scientific bases for subsequent decisions. We demonstrate an integration of monitoring data with a dynamic, metapopulation model developed to assess effects of streamflow alteration on fish occupancy in a southeastern US stream system. Although not extensive (collected over three years at nine sites), the monitoring data allowed us to assess and update support for alternative population dynamic models using model probabilities and Bayes rule. We then use the updated model weights to estimate the effects of water withdrawal on stream fish communities and demonstrate how feedback in the form of monitoring data can be used to improve water resource decision making. We conclude that investment in more strategic monitoring, guided by a priori model predictions under alternative hypotheses and an adaptive sampling design, could substantially improve the information available to guide decision-making and management for ecosystem services from lotic systems.