Food webs are not static over time, but our knowledge on their dynamics is extremely scarce due to methodological challenges. These have imposed significant limitations on our ability to mechanistically understand how temporal changes affect trophic networks. Here, we address this gap using high-throughput molecular diagnostics to measure the season-wide dynamics of trophic interactions between invertebrate generalist predators, pest and alternative prey in replicated cereal fields across two years. We used the level of food web specialisation as a proxy for predator redundancy in pest control, and hypothesized that it would hit its minimum, and invertebrate diversity its maximum, at the middle point of the season. Additionally, alternative prey availability was indirectly increased by adding manure to half of each field, to test if this would reduce specialisation. In line with our predictions, it showed an inverse bell-shaped curve over the season, while prey, but not predator, diversity showed an opposite trend. No significant effects of fertilization were found on food web specialisation. Our findings identify early and late season as the times when generalist predators are behaviourally most constrained, pin-pointing these periods as the ones with the lowest redundancy in pest control. Hence, molecular trophic analyses provide unique insights into the temporal dynamics of food webs and their properties. This allows the generation of temporal roadmaps for when management interventions are expected to be most effective.