fundamentals of eDNA

It was interesting to read about the earlier days of eDNA in the deSouza et al. paper. Even today, seasonality is not given as much consideration as it should deserve in many eDNA studies, and seasonality itself can be highly organism- and location-specific.

Nevertheless, the paper had several flaws that must be considered when it comes to project design for our own theses. Seasonality was poorly defined in this study, and seasonality in turn has effects on differences in environmental physicochemical parameters that affect the ecology of eDNA, such as tannin inhibition from leaves in fall, and differences in flow rates across the seasons. Is there a distinct temperature threshold where biological activity is altered? It is critical to make use of appropriate environmental and biological data when making such broad, vague definitions.

While less well understood at the time of the study, the ecology of one’s study organism cannot be viewed in isolation from the ecology of eDNA, which must be constrained as best as possible by accounting for confounding variables. At the same time, one must have a strong understanding of the behaviour and ecology of one’s target species in order to design a good study, although this observation is less relevant for my own future work on relatively poorly-understood microbial and meiofunal communities.

It is also critical to be consistent when it comes to sampling methodology, although I am of the opinion that any usable related data should not be ‘wasted’ even if it was conducted under a different methodology, with caveats for their incorporation into analyses.

In this study, while the lab-based methodologies were fairly well documented, there was a dearth of information with regards to field methodologies. It is important to be thorough when documenting all steps of the research process, for accountability, QC and reproducibility, among other factors. Given the relative ease of measuring physicochemical parameters these days (e.g. with sondes), we should collect as much of such data as feasible, instead of making too many assumptions about physical environmental parameters as in this study (e.g. not directly measuring flow rate, but perhaps using slope as a proxy for flow rate).

As best as possible, in our own work, we should optimise our study designs and data analyses to fit our research questions, although it is highly likely that we have to make use of external datasets that are optimised for previous studies, and approaches that aren’t optimised for any specific study, such as the index sites data. We need to make sure that the spatial and temporal scales of the external datasets we do use is appropriate for our work at hand.

Finally, Andy brought up a salient point that being overly thorough may not be practical in some cases. We do our due diligence, and take the data and observations we have to convey something useful to the scientific (and wider) community. There’s no need to chase too tightly after perfection.