Water temperature plays a significant role in determining the interaction strengths between marine fish species, and thus plays a critical role in controlling community dynamics in coastal ecosystems. However, it is challenging to track and quantify these interactions under field conditions. To address this, we performed quantitative fish environmental DNA (eDNA) metabarcoding on 550 seawater samples collected bimonthly from 11 coastal sites over two years in the Boso Peninsula, Japan. We analyzed eDNA monitoring data using nonlinear time series analytical tools, detecting fish-fish interactions as information flows between eDNA time series. We reconstructed interaction networks for the top 50 frequently detected species and quantified pairwise, fluctuating interaction strengths. We found that water temperature influenced fish-fish interaction strengths, but the impact varied among fish species. This study demonstrates a practical research framework to study the effects of environmental variables on interaction strengths of marine communities in nature, which would contribute to understanding and predicting natural marine ecosystem dynamics. An increase in water temperature induced by global climate change may change fish interactions in a complex way, which consequently influences marine community dynamics and stability.