Salish Sea Ecosystem Conference

Using Bayesian networks to predict water quality based on land use and climate stressors in Australia as a model for application to the Salish Sea.

2016SSEC

Bayesian Networks (BNs) are a useful tool for natural resource management and decision-making. In this case study, they are used in conjunction with the Relative Risk Model to assess ecological risk to estuaries in Southeast Queensland, Australia (SEQ). Similar to the Salish Sea, nutrient and sediment loading from anthropogenic development are major stressors to water quality and biota in SEQ. The combination of runoff from heavy rainfall, excess nutrients, phytoplankton blooms and a large supply of organic matter cause eutrophication and changes to aquatic communities. Furthermore, global climate change may be an important factor to consider for predicting future water quality. Water quality objectives have been developed to assess the condition of estuaries and guide management. We have built risk assessment models using BNs for three estuaries in SEQ to estimate the probability of meeting those objectives in different regions of the estuaries. The relationships in the model were determined using case-learning with over 10 years of monthly monitoring data. I have also integrated the response of eukaryotic benthic communities to water quality. The benthic community data were collected via DNA metabarcoding where DNA is collected from environmental media (such as soil, sediment or water), sequenced, and identified using online databases. I will be presenting the methodology of building the models, current findings, and lessons learned for using similar tools in the long-term management of the Salish Sea, even under climate change. Similar probabilistic models could be built based on monitoring data from the Salish Sea, regional management objectives in the Pacific northwest, and climate projections.