3. Limitations and perspectives
We noted some potential biases and limitations of the dataset used in our meta-analyses. Firstly, studies estimating the decay rates of nuclear eDNA were substantially fewer when compared with those on mitochondrial eDNA, particularly in freshwater systems (Figure 3), which might limit our ability to infer the effect of water source on eDNA degradation between the target genes. In addition, eDNA decay rates targeting longer DNA fragments (>200 bp) and taxa other than fish were relatively scarce. Moreover, estimation of eDNA decay rates using a 0.7 µm pore size filter appeared to be relatively more common, which suggests greater knowledge of eDNA persistence in this filter pore size, and a potential bias in our meta-analyses. It is expected that eDNA analysis will be applied to ecological monitoring of more varied taxa and environments in the future, and will have to be developed accordingly to determine the spatiotemporal scale of eDNA signals and to maximize the biological information obtained from eDNA samples. More information on eDNA persistence and degradation should therefore be collected, by targeting different taxa and environments and using various collection and analysis methods.
Although our findings and their implications require further verification, this study is the first to propose that the persistence of eDNA from macro-organisms can be determined by the state of the eDNA and its complex interactions with environmental conditions, i.e. the mechanism of eDNA persistence and degradation cannot be fully understood without knowing not only the environmental biotic and abiotic factors involved in eDNA degradation but also the cellular and molecular states of eDNA occurring in water. If our findings are correct, the spatiotemporal scale and intensity of eDNA signals would be different depending on the eDNA particle size and state. The fact that Weibull or biphasic exponential decay models fit better to eDNA degradation implies the differences in eDNA persistence depending its state (e.g., intra- or extra-cellular, living or dead cells, particulate or dissolved) (Eichmiller et al., 2016; Bylemans et al., 2018), which support our results linking eDNA persistence to its state. In addition, the study by Jo et al. (2020c), where it was reported that the genomic information obtained from eDNA samples can differ depending on the filter pore size, can further support the link between eDNA state and persistence. Experimental verification of our findings and implications will highlight the importance of clarifying the characteristics and dynamics of aqueous eDNA, and will contribute substantially to the development of eDNA analysis in the future.