Development of protocols for environmental impact studies using causal modelling, with special reference to wastewater discharges to freshwater systems

Submission note: A thesis submitted in total fulfilment of the requirements for the degree of Doctor of Philosophy to the Department of Ecology, Environment and Evolution, La Trobe University, Bundoora.

As the population of the world is increasing and water resources have become scarce, an appropriate assessment of freshwater systems has become increasingly important. A major challenge to assessing the impacts of human activities is accounting for natural spatiotemporal variation. Current freshwater assessments are unable to solve spatiotemporal confounding through study designs or statistical analyses. Study designs in observational studies fail to consider the period-by-location interaction in the absence of an impact and using statistical analysis alone is inadequate. In the present study, the problem of spatiotemporal confounding was addressed using causal modelling based on spatiotemporal data to infer causal effects of wastewater on biotic ecosystems. A combination of statistical analysis and the theory of causation was used to address confounding bias and to predict the effect of future interventions. Benthic macroinvertebrate and environmental variable data were collected from eight sites upstream and downstream of a wastewater treatment plant discharge point along 8km of waterway at five times of sampling times over 1.5 years. The composite hypotheses based on the theoretical relationships among these variables were summarised in a causal diagram. Model building and testing was conducted using Structural Equation Modelling (SEM). Distance-based redundancy analysis (dbRDA) was used for model building and hypothesis testing. The results indicated that the causal effects of effluent on macroinvertebrate communities could be inferred using causal modelling. The macroinvertebrate communities responded to water quality degradation with a clear shift in community composition after the discharge point that varied seasonally. Chlorophyll a, TOC, zinc, conductivity, temperature and its interaction with conductivity were important determinants of the macroinvertebrates. Causal models also explained the spatiotemporal variations in environmental variables. The consistency of data with the structure of the causal diagram was confirmed with global Fisher’s C-test. The final causal models were used to predict the effect that manipulating the values of observed discharge would have on abiotic variables; demonstrating how causal modelling can be used for developing protocols for environmental impact studies.