LWRRDC Final Report: Temporal variability of macroinvertebrate communities in Australian streams
Internal Report 298
Supervising Scientist Division
About the report
The degree and extent of temporal variability of stream macro invertebrate communities have been investigated across a broad cross-section of climatic/hydrological regimes in Australia. Constancy or persistence of macro invertebrate communities was found to be significantly and positively correlated 'with permanence of stream flow, and negatively correlated with interannual variability of annual stream discharge. There was a tendency (only) for macro invertebrate communities of permanent streams in temperate Australia to be more persistent than those in tropical regions. Temporal variability is believed to have most potential to limit AUSRIVAS sensitivity and to result in greater model output failures for sites in northern Australia (QLD inclusive) and possibly for sites in drought-prone portions of warm-temperate, eastern Australia. Drought in eastern Australia and major disturbance arising from cyclones in northern Australia appear to be the major contributors to high temporal variability of macroinvertebrate communities.
Whilst a preliminary study was undertaken to determine the consequences to model development (classification) of temporal variability, a more complete sensitivity analysis is required in future MRHI R&D to determine the full implications of collective error and variability (at various spatial scales) for model sensitivity. This analysis would include the determination of the sizes of various sources of error and variation and their effects on the rates of misclassification to quality bands. Data on temporal variability arising from the current study will provide an important information base upon which such an assessment can proceed. Future R&D needs that will assist in this 'sensitivity analysis' have been identified in the report and attachments.
The study formulated a number of approaches to pursue in relation to temporal variability and predictive modelling. This included approaches for assessing implications for predictive model sensitivity arising from temporal variability, as well as approaches that might be used to account for such variability, ie: (i) contextual data for assessing the severity of temporal variability, (ii) modelling temporal variability, (iii) adjusting and updating model output, (iv) models for different climatic conditions, and (v) combined-seasons or -years models. Whilst at this stage the extent to which high temporal variability may compromise the sensitivity of predictive models is not known, the ability to reliably identify and predict different geographical regions and stream types susceptible to high temporal variability are in themselves informative and valuable for management. The magnitude of persistence indices calculated in this study and modelled according to different regions and stream types, may eventually be related to some measure of AUSRIVAS model 'noise' and variability and, consequently, to poor model predictions. With quantified degrees of 'risk' of model failure, researchers and managers might then be better informed and placed to account for such variability, stipulate error and probability statements around predictions, or recommend alternative monitoring approaches.