3. Re-analysis of the time-series changes in eDNA particle size distribution
To assess the validity of the findings of the meta-analyses, we re-analysed the dataset from a previous study investigating the particle size distribution of eDNA derived from the mitochondria and nuclei of Japanese jack mackerel (Trachurus japonicus ) and the time-series changes therein, after fish removal from tanks (Jo et al., 2019b). In the aforementioned study, mitochondrial and nuclear eDNA degradation was examined under multiple size fractions, and both degradations tended to be suppressed at smaller size fractions. We estimated the eDNA decay rate constants at different size fractions using the dataset from the said study, and assessed the variation in eDNA decay rates depending on the eDNA particle size, target gene, and water temperature. Detailed information on the experimental design, water sampling, and molecular analyses can be found in Jo et al. (2019b).
We included all eDNA samples that could pass through sequential filters with 10, 3, 0.8, and 0.2 μm pore sizes at 0, 6, 12, and 18 hours, which yielded four eDNA size fractions, i.e. >10, 3-10, 0.8-3, and 0.2-0.8 µm. Linear regressions were performed between eDNA concentrations (original concentration + 1 followed by log-transformation) and sampling time points for each size fraction, target gene (mitochondrial or nuclear), and temperature level (13, 18, 23, or 28 °C), to estimate the slope (i.e. eDNA decay rate constant) and the corresponding 95% CI, using lm and confint functions in R, respectively. Here, the two fish biomass levels (Small and Large; see Jo et al. (2019b)) were pooled to increase the sample size. We then performed ANOVA to assess the relationship between eDNA degradation, particle size, target gene, and temperature. We included the median of the slope (eDNA decay rate) as the dependent variable, and the filter pore size, target gene, water temperature, and their primary interactions as the explanatory factors.