Bioindicator monitoring and modelling for informing river health management

Abstract

Freshwaters are considered to be the most degraded ecosystems. In an attempt to monitor river health, macroinvertebrates and diatoms proved to be suitable for routine short- and long-term monitoring. However, making complex decisions for river health management based solely on the data generated by such monitoring efforts is challenging, mainly when data has limited predictive capacity. To overcome this, the thesis focuses on the integration of bioindicator monitoring and novel modelling techniques with the associated challenges to inform river health management. This thesis aimed firstly to document differences between two local catchments using ecological threshold models and then test the performance of two commonly used models. Results suggested three times higher total phosphorus (TP) thresholds in the Onkaparinga River catchment as compared to the River Torrens catchment. It was also found that thresholds for electrical conductivity (EC) specified by Hybrid Evolutionary Algorithms (HEA) exceeded those identified by Threshold Indicator Taxa Analysis (TITAN). Importantly, despite the observed differences in thresholds, results indicated that South Australian water quality guidelines for freshwater systems might be too high. Another study aimed to identify model-based bias in threshold identification, found that two commonly used methods, i.e. gradient forest (GF) and TITAN are robust in identifying change in species responses. Still, threshold identification differs depending on the analysis used and the nature of ecological data. Noting the differences in performance of the two methods on the same field and synthetic data, we recommend careful application of GF and TITAN, will improve their use for river health management. Another critical and neglected aspect of threshold identification is dependence on spatial scale. Data analysis from field monitoring of diatoms and water quality during autumn and spring for two years was further extended by merging with a broader data set from South Australian streams, to highlight spatial influences on thresholds. Consistent or lower thresholds were found across spatial scales when spatial resolution was increased from state to local scale. However, higher TP and EC thresholds were observed for the South East region than for the Adelaide and Mount Lofty Ranges. Thus we highlighted that thresholds derived at broad scales alone are unlikely to be appropriate for finer-scale assessment. Another study applied an integrated modelling approach using GF, Soil and Water Assessment Tool (SWAT) and HEA to predict river health for future climate and land use. This study quantified population dynamics of sensitive taxa identified by GF under SWAT simulated future scenarios of deforestation, reforestation, climate change, and 10% increased urbanisation, as projected by local authorities over the next 30 years. HEA, used to predict future abundances, suggested a nonlinear response of species within commonly used Ephemeroptera, Plecoptera and Trichoptera grouping and we recommend to redirect focus of such studies from community to species-level. In reforestation and climate change scenarios, SWAT results also suggested a shift in the unusual permanent flowing stream of Sixth Creek towards intermittency under climate change and reforestation scenarios. Overall, outcomes of this research provide an improved understanding of local catchment processes in the context of global-scale challenges of climate and land use changes.