On-site observation of reservoir EPA/DHA spatiotemporal distribution using a multi-wavelength excitation fluorometer and machine learning

Aoki N, Fujibayashic M, Sakamaki T, Kuba T. 2026. On-site observation of reservoir EPA/DHA spatiotemporal distribution using a multi-wavelength excitation fluorometer and machine learning. Lake Reserv Manag. XXX–XXX.The quality of food resources strongly influences the health of aquatic animals in lacustrine environments. Eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) are fatty acids that are indicators of food quality. The content (absolute mass per unit lake water volume) of these fatty acids can be used to assess the adequacy of essential biochemicals in lacustrine food webs. Their levels (mass percentage of total fatty acids in suspended solids) reflect the nutritional balance and trophic transfer efficiency to higher trophic levels. Their distributions vary spatially and temporally, and detailed dynamics have rarely been captured because conventional analyses of EPA and DHA are time-consuming and labor-intensive. In this study, we applied a recently developed, simple observation method that combines a multi-wavelength excitation fluorometer with machine learning to monitor EPA and DHA contents and EPA and DHA levels. This method was used to measure these acids monthly in Hokuzan Reservoir, Saga Prefecture, Japan, from July 2021 to December 2023. Distinct vertical distribution patterns and seasonal periodicity were observed. During the summer, EPA and DHA were concentrated in the epilimnion attributed to thermal stratification, whereas winter turnover resulted in uniform and elevated concentrations throughout the water column. Seasonal variations in EPA and DHA levels indicated compositional shifts dominated by EPA- and DHA-rich phytoplankton species. The approach demonstrated here offers a practical framework for monitoring the nutritional quality of aquatic food webs and for managing lake ecosystems.