Eukaryotic harmful and toxic microalgae, along with their derived toxins, pose significant threats to seafood safety, human health, and marine ecosystems. Here, we developed a novel full-length 18S rRNA database for harmful and toxic microalgae and combined metabarcoding with toxin analyses to investigate the ecological patterns of phytoplankton communities and the underlying mechanism of associated toxic microalgae risks. We identified 79 harmful and toxic species in Hong Kong's coastal waters, with dinoflagellates and diatoms representing the majority of toxic and harmful taxa, respectively. Distinct seasonal succession patterns were observed in phytoplankton communities, driven by different ecological assembly processes. Deterministic processes dominated during the dry season, correlating with elevated toxic microalgae abundance and temperature stress. Seasonal shifts in temperature played a pivotal role in shaping toxic algal communities. The dominance of dinoflagellates, particularly spp., spp., spp., and spp., during the dry season was consistent with elevated toxin concentrations. These toxin profiles highlight the heightened risk in a warming climate, where the prevalence and impacts of toxigenic algae are expected to intensify.