Josh Griffiths1, Tamielle Brunt2, Dr Luke Noble1
1EnviroDNA, Brunswick, Melbourne, Australia, 2Wildlife Queensland , Brisbane, Australia
Biography:
Josh is a Wildlife Ecologist with over 20 years experience in designing and implementing programsto improve our understanding of Australia’s native fauna using traditional, and now eDNAtechniques. For the past 15 years, Josh has specialized in platypus research and is recognized asone of Australia’s leading platypus ecologists.Josh has designed and managed a diverse array of projects including assessing the impacts ofenvironmental disturbances, identifying variables that influence spatial and temporal populationtrends, understanding the response to management actions, and broad-scale, citizen science ledbaseline monitoring programs. This has included the novel application and development of newtechniques such as acoustic telemetry, eDNA, and platypusSPOT.
Abstract:
Freshwater ecosystem management depends on indicators that can be measured consistently, interpreted against agreed references, and applied at the density required for catchment-scale decisions. In South East Queensland, the Ecosystem Health Monitoring Program provides a long-term record of biotic and abiotic indicators from extensive surveys. Programs of this kind are critical but costly, limiting assessment frequency and scale, and broader adoption outside flagship programs. We examined how eDNA tracks established indicators in the region, and what properties a molecular indicator of condition requires.
Large-volume water samples were collected across the region, analysed by metabarcoding, and detections compared with co-located survey records. Indicators were derived by substituting detected taxa into existing indices, by supervised models trained against the survey indicators, and by unsupervised one-dimensional community indices. Predictive performance was assessed under cross-validation designs matched to distinct intended uses, including interpolation within the monitored network and extrapolation to unmonitored catchments and to under-represented habitat types.
Molecular indicators generally recovered the established indicators, with performance varying across indicators and the validation design. Transfer was weaker to unsampled catchments and weakest to distinct habitat types. Community-level data was more powerful than summary indices. Unsupervised indices assume the dominant compositional gradient corresponds to condition, an assumption not supported within this region and sampling structure.
Molecular indicators are most appropriately framed as reference-based and stratified by habitat type, and reported with an explicit domain of applicability and quantified uncertainty. In this framework, eDNA can provide cost-effective, multi-use data that complements existing programs and facilitates uptake.