The use of Bayesian networks to guide investments in flow and catchment restoration for impaired river ecosystems

1. The provision of environmental flows and the removal of barriers to water flow are high priorities for restoration where changes to flow regimes have caused degradation of riverine ecosystems. Nevertheless, flow regulation is often accompanied by changes in catchment and riparian land-use, which also can have major impacts on river health via local habitat degradation or modification of stream energy regimes.

2. The challenges are determining the relative importance of flow, land-use and other impacts as well as deciding where to focus restoration effort. As a consequence, flow, catchment and riparian restoration efforts are often addressed in isolation. River managers need decision support tools to assess which flow and catchment interventions are most likely to succeed and, importantly, which are cost-effective.

3. Bayesian networks (BNs) can be used as a decision support tool for considering the influence of multiple stressors on aquatic ecosystems and the relative benefits of various restoration options. We provide simple illustrative examples of how BNs can address specific river restoration goals and assist with the prioritisation of flow and catchment restoration options. This includes the use of cost and utility functions to assist decision makers in their choice of potential management interventions.

4. A BN approach facilitates the development of conceptual models of likely cause and effect relationships between flow regime, land-use and river conditions and provides an interactive tool to explore the relative benefits of various restoration options. When combined with information on the costs and expected benefits of intervention, one can derive recommendations about the best restoration option to adopt given the network structure and the associated cost and utility functions.