Marine ecosystem models often consider temporal dynamics on the order of months to years, and spatial dynamics over regional and global scales as a means to understand the ecology, evolution, and biogeochemical impacts of marine life. Large-scale dynamics are themselves driven over diel scales as a result of light-driven forcing, feedback, and interactions. Motivated by high-frequency measurements taken by Lagrangian sampling in the North Pacific Subtropical Gyre, we develop a hierarchical set of multitrophic community ecology models to investigate and understand daily ecological dynamics in the near-surface ocean including impacts of light-driven growth, infection, grazing, and phytoplankton size structure. Using these models, we investigate the relative impacts of viral-induced and grazing mortality for Prochlorococcus; and more broadly compare in silico dynamics with in situ observations. Via model-data fitting, we show that a multi-trophic model with size structure can jointly explain diel changes in cyanobacterial abundances, cyanobacterial size structure, viral abundance, viral infection rates, and grazer abundances. In doing so, we find that a significant component (between 5% to 55%) of estimated Prochlorococcus mortality is not attributed to either viral lysis (by T4- or T7-like cyanophage) or grazing by heterotrophic nanoflagellates. Instead, model-data integration suggests a heightened ecological relevance of other mortality mechanisms -- including grazing by other predators, particle aggregation, and stress-induced loss mechanisms. Altogether, linking mechanistic multitrophic models with high-resolution measurements provides a route for understanding of diel origins of large-scale marine microbial community and ecosystem dynamics.