Enabling Quantitative eDNA Monitoring in Lotic Ecosystems by Incorporating Hydrological Characteristics and Allometric Scaling

Abstract

Effective freshwater biodiversity monitoring requires scalable methods to estimate fish abundance or biomass. Environmental DNA (eDNA)‐based approaches offer a promising alternative to observational methods. However, translating eDNA concentrations into reliable biomass estimates remains difficult in flowing waters because hydrological dilution and transport weaken and spatially displace eDNA signals. As a result, the eDNA to biomass or abundance relationship is often highly variable and insufficient for quantitative monitoring. We examined species‐specific relationships between eDNA signals, fish abundance, and biomass using mesocosm experiments and landscape‐scale field surveys in lotic environments and compared the performance of targeted barcoding and metabarcoding approaches for five species. We first quantified these relationships under controlled conditions and then tested whether incorporating river discharge rate, habitat complexity, and species‐specific traits could improve predictions in natural systems. In mesocosms, droplet digital PCR (ddPCR) assays revealed strong correlations between eDNA concentrations and both fish abundance and biomass, e.g., Barbatula barbatula ( r 2  = 0.91, p  < 0.001) for abundance, Rutilus rutilus ( r 2  = 0.79, p  < 0.05) and Cottus perifretum ( r 2  = 0.76, p  < 0.001) for biomass. The relative strength of abundance‐ versus biomass‐eDNA relationships depended on the degree of individual size variation within each species. Field surveys identified river discharge rate and species‐specific traits as strongest correlates of eDNA concentration, while habitat complexity was less influential. A linear mixed‐effects model confirmed ddPCR's reliability for species quantification ( r 2  = 0.29 for abundance and r 2  = 0.16 for biomass). Metabarcoding yielded lower model concordance ( r 2  = 0.24 for abundance and r 2  = 0.03 for biomass), suggesting more limited quantitative accuracy compared to ddPCR. Our results show that integrating discharge‐derived hydrological characteristics substantially strengthens the eDNA to biomass or abundance relationship, enabling a predictive framework for estimating fish biomass in dynamic lotic systems.