Oral Presentation Freshwater Sciences 2023

Prediction of stream macroinvertebrate assemblage responses to human impacts is improved by DNA metabarcoding methods (#460)

Thomas E Wilkins 1 , Chris J Walsh 1 , Yung E Chee 1 , Melissa E Carew 1 2 , Ary A Hoffmann 2 , Rhys A Coleman 1 3
  1. Waterway Ecosystem Research Group, University of Melbourne, Burnley, Victoria, Australia
  2. Pest & Environmental Adaptation Research Group, University of Melbourne, Parkville, Victoria, Australia
  3. Melbourne Water Corporation, Docklands, Victoria, Australia

Traditional morphological identification of in-stream macroinvertebrates as part of routine biological monitoring programs can be costly and often limited to family-level taxonomic resolution. DNA metabarcoding can provide cost effective and accurate species-level identification and potentially allow us to better quantify species richness and better understand taxon-specific responses to environmental gradients and human impacts. We analysed a stream macroinvertebrate dataset designed to sample a range of environmental and human impact gradients at 46 streams across greater Melbourne. We parsed the dataset into: i) morphological family-level presence-absence data, ii) morphological family-level abundance data, iii) DNA-metabarcoded family-level presence-absence data and iv) DNA-metabarcoded species-level presence-absence data. With these, we developed four hierarchical multi-taxon models to quantify taxon relationships to catchment area, mean annual runoff, imperviousness (I), forest cover (F) and the interaction between imperviousness and forest cover (I:F).

There were ~100 taxa in the three family-level models and 617 species in the species-level model. In other words, DNA-metabarcoded species-level data revealed six times as many “taxonomic units” as family-level data across our 46 sites. Morphological and DNA-barcoded family-level presence-absence models both estimated similar taxon responses to our predictors of interest. Morphological family-level abundance data did not improve model fit or predictive performance over morphological family-level presence-absence data. We also found more non-zero taxon responses to predictors I, F and I:F with DNA-metabarcoded species-level data than with DNA-metabarcoded family-level data.

DNA metabarcoding provides more accurate estimates of species richness which has not previously been possible, due to the constraints of costs and expertise necessary for species-level morphological identification. It also gives us the ability to investigate animal-environment relationships at a more granular taxonomic level. DNA-metabarcoded data from sorted samples can provide equivalent information to morphologically-sorted family-level data thus providing comparability with historical data.