Marine autotrophic plankton communities are fundamental components of coastal ecosystems, supporting marine food webs and contributing to ecosystem functioning through primary production. Understanding how their taxonomic composition and size structure vary over time is essential for assessing ecosystem responses to environmental changes. In this study, we used a size-fractionated eDNA metabarcoding approach to investigate phytoplankton community dynamics in a temperate coastal area of the northwestern Adriatic Sea over a five-year period (2020–2024). Seawater samples were separated into three size fractions (3–10 μm, 10–50 μm, and 50–100 μm), representing different phytoplankton functional groups, and analyzed by high-throughput sequencing of the 18S rRNA V4 region. Overall, Alveolata and Stramenopiles phyla dominated the communities. Strong seasonal patterns emerged. Winter communities were characterized by higher chlorophyll-a concentrations (1.15 ± 0.34 μg L⁻¹), greater diversity under nutrient-rich mixed-water conditions, whereas summer assemblages showed lower biomass (0.22 ± 0.11 μg L⁻¹) and a more homogeneous composition under stratified, nutrient-poor conditions. Stramenopiles dominated winter communities in 2020–2021, while a shift toward higher Alveolata abundance was observed in subsequent years. Community composition was mainly driven by size fraction (R² = 0.337; F = 5.51, p = 0.001), with season exerting a secondary influence (R² = 0.092). Beta diversity analyses revealed that species turnover accounted for most compositional variation (88–92%), indicating that seasonal changes were driven primarily by species replacement rather than nestedness. Despite marked taxonomic and phylogenetic shifts, biomass distribution among size classes remained relatively stable, suggesting that functional organization is more conserved than species composition. These findings highlight the value of size-fractionated eDNA metabarcoding for understanding phytoplankton dynamics and suggest that functional stability may help maintain ecosystem functioning despite substantial taxonomic turnover.