Estimating fish population abundance by integrating quantitative data on environmental DNA and hydrodynamic modelling

Molecular analysis of DNA left in the environment, known as environmental DNA (eDNA), has proven to be a powerful and cost-effective approach to infer occurrence of species. Nonetheless, relating measurements of eDNA concentration to population abundance remains difficult because detailed knowledge on the processes that govern spatial and temporal distribution of eDNA should be integrated to reconstruct the underlying distribution and abundance of a target species. In this study, we propose a general framework of abundance estimation for aquatic systems on the basis of spatially replicated measurements of eDNA. The proposed method explicitly accounts for production, transport and degradation of eDNA by utilizing numerical hydrodynamic models that can simulate the distribution of eDNA concentrations within an aquatic area. It turns out that, under certain assumptions, population abundance can be estimated via a Bayesian inference of a generalized linear model. Application to a Japanese jack mackerel (Trachurus japonicus) population in Maizuru Bay revealed that the proposed method gives an estimate of population abundance comparable to that of a quantitative echo sounder method. Furthermore, the method successfully identified a source of exogenous input of eDNA (a fish market), which may render a quantitative application of eDNA difficult to interpret unless its effect is taken into account. These findings indicate the ability of eDNA to reliably reflect population abundance of aquatic macroorganisms; when the ecology of eDNA is adequately accounted for, population abundance can be quantified on the basis of measurements of eDNA concentration.

Automated summary

The study demonstrates that environmental DNA (eDNA) analysis can be a powerful tool to estimate population abundance of aquatic macroorganisms. By integrating knowledge of the processes governing eDNA distribution with numerical hydrodynamic models, a Bayesian inference of a generalized linear model can be used to estimate population abundance. The method was shown to provide comparable population abundance estimates to traditional methods and can also identify sources of exogenous eDNA input.