Estimation accuracy and eDNA assay strategy
In addition to eDNA production source and state, we investigated the effects of abundance metrics and eDNA assay strategy, including collection and quantification strategies, on the estimation accuracy of species abundance via eDNA. First, there was no significant difference in the estimation accuracy between biomass and number of individuals. This is consistent with the finding in Yates et al. (2019) although our study meta-analyzed a larger number of studies. However, given that wild populations are often composed of individuals with different age classes and body sizes, selecting either metrics itself may ultimately hamper the accurate estimation abundance to select either of abundance metrics. A few studies recently applied allometric scaling coefficients to the relationship between eDNA quantity and species abundance, showing higher R2 values for allometrically scaled mass (ASM) than biomass and individuals (Chin et al., 2021; Yates et al., 2021a; 2021b).
Second, the estimation accuracy of species abundance was not affected by the assay strategy (filtration/centrifugation and species-specific/metabarcoding). In particular, although the variation of R2 values were higher for metabarcoding than species-specific assay, it is surprising that metabarcoding assay had similar abundance estimation accuracy to species-specific assay (real-time or digital PCR). However, the result must be tempered with an acknowledgement that (i) the proportion of studies conducting laboratory experiments were higher for metabarcoding assay, (ii) there was no result of metabarcoding assay targeting crustaceans eDNA, and (iii) the number of literatures performing metabarcoding assay was much fewer (9 out of 56) than that of species-specific assay (Tables 1 and S1). Theoretically, as the number of eDNA reads is estimated from the PCR final product, metabarcoding assay should not be suitable for the accurate estimation of species abundance, except for the application of internal standard DNAs (i.e., known amounts of short artificial DNA fragments; Ushio et al., 2018). In addition, regardless of the substantial difference in the volume of water sample, there was no difference in the estimation accuracy between filtration and centrifugation. This implies that water sample volume does not have a substantial effect on the estimation accuracy, partially contradicting the above hypothesis that larger volume of water samples might allow to improve the estimation accuracy of species abundance. Further studies are required how the assay strategy and sampling water volume relate to the accuracy of eDNA-based species abundance estimation.