A new simulation framework to evaluate the suitability of eDNA for marine and aquatic Environmental Impact Assessments

A model was developed to forecast and compare changes in species presence assessed with either eDNA or traditional observations. We use it to explore how ecosystem conditions could affect the suitability of eDNA for Environmental Impact Assessment. First, a deterministic model simulated the dynamics of the impacted population (called “receptor” in EIA) and their shed DNA fragment concentrations. Second, random distributions of receptor organisms and eDNA fragment quantities at steady state were simulated within the impacted spatial domain (called “project area”). Then, simple random samplings were performed for both the receptor and eDNA. Third, post-sampling processes (eDNA extraction, amplification, and analysis) were simulated to estimate the taxon detection probability. Fourth, we simulated an impact by modifying the growth, mortality, and mobility (null, passive, and active) parameters of the receptor taxon. eDNA detection probability curves were then estimated for a range of environmental sample volumes by fitting a Weibull cumulative distribution function. A F-like statistic compared detection curves before and after impact. Statistically significant differences were detected with eDNA in impact scenarios where receptor taxon growth rate decreased and receptor mobility was null or passive. In scenarios where the project area accumulates DNA shed from multiple categories of the same taxon (e.g., from dead organisms if mortality increased or when individuals can cross project area boundaries), it is difficult to assess impact. Our study shows that results obtained from eDNA sampling will not always agree with an impact classically assessed on a receptor population. One reason is that sources of the total eDNA pool are not identified. The modeling highlights the need: to do preliminary testing of sample sizes, to develop new approaches that will identify sources from the pool of extracted DNA, and to improve descriptions of the ecogeochemical processes required to forecast shed DNA reactivity.