Using Bayesian networks to predict risk to estuary water quality and patterns of benthic environmental DNA in Queensland

• Published in: Integrated environmental assessment and management 2019-01, Vol.15 (1), p.93-111
• Main Authors: Graham, Scarlett E, Chariton, Anthony A, Landis, Wayne G
• Format: Article
• Language: English
• Subjects: Adaptive management; Algae; Animals; Aquatic Organisms - genetics; Aquatic plants; Bayesian analysis; Bayesian network; Benthos; Biodiversity; Chlorophyll; Chlorophyll a; Communities; Community structure; Deoxyribonucleic acid; Diatoms; DNA; DNA sequencing; DNA structure; Ecological risk assessment; Environmental DNA; Environmental DNA (eDNA); Environmental Monitoring - methods; Estuaries; Estuarine dynamics; Estuarine environments; Eukaryotes; Fungi; Heterotrophic organisms; Heterotrophs; Invertebrates - genetics; Mathematical models; Modelling; Natural resource management; Natural resources; Pine; Prediction models; Probability theory; Queensland; Regions; Resource management; Risk assessment; Water Pollution - statistics & numerical data; Water Quality
• ISSN: 1551-3777; 1551-3793
• DOI: 10.1002/ieam.4091

ABSTRACT Predictive modeling can inform natural resource management by representing stressor–response pathways in a logical way and quantifying the effects on selected endpoints. This study demonstrates a risk assessment model using the Bayesian network relative risk model (BN‐RRM) approach to predict water quality and, for the first time, eukaryote environmental DNA (eDNA) data as a measure of benthic community structure. Environmental DNA sampling is a technique for biodiversity measurements that involves extracting DNA from environmental samples, amplicon sequencing a targeted gene, in this case the 18s rDNA gene (which targets eukaryotes), and matching the sequences to organisms. Using a network of probability distributions, the BN‐RRM model predicts risk to water quality objectives and the relative richness of benthic taxa groups in the Noosa, Pine, and Logan estuaries in Southeast Queensland (SEQ), Australia. The model predicts Dissloved Oxygen more accurately than the chlorophyll a water quality endpoint and photosynthesizing benthos more accurately than heterotrophs. Results of BN‐RRM modeling given current inputs indicate that the water quality and benthic assemblages of the Noosa are relatively homogenous across all sub risk regions, and that the Noosa has a 73%–92% probability of achieving water quality objectives, indicating a low relative risk. Conversely, the Middle Logan, Middle Pine, and Lower Pine regions are much less likely to meet objectives (15%–55% probability), indicating a relatively higher risk to water quality in those regions. The benthic community richness patterns associated with risk in the Noosa are high Diatom relative richness and low Green Algae relative richness. The only benthic pattern consistently associated with the relatively higher risk to water quality is high richness of fungi species. The BN‐RRM model provides a basis for future predictions and adaptive management at the direction of resource managers. Integr Environ Assess Manag 2019;15:93–111. © 2018 SETAC Key Points We demonstrated that it is possible to use case learning to build a Bayesian network by using the Bayesian network relative risk model to predict the probability of meeting water quality objectives for estuarine regions. It is also possible to use case learning to predict the patterns of occurrence of key eukaryotic taxonomic groups by using environmental DNA (eDNA) as the identifier. Once the Bayesian network has been built and the predictions cro