The plant stress and plant vigor hypotheses are widely used to explain the distribution and abundance of insect herbivores across their host plants. These hypotheses are the subject of contentious debate within the plant herbivore research community, with several studies finding simultaneous support for both hypotheses for the same plant–herbivore interaction. We address the question of how such support is possible using dynamic site‐occupancy models to quantify the attack dynamics of Cryptorhynchus lapathi (poplar‐willow weevil) on Salix sitchensis (Sitka willow), a dioecious shrub colonizing Mount St. Helens, Washington, USA after the 1980 eruption, in relation to host plant stress, vigor, and sex. We also introduce several scaling criteria as a rigorous test of the plant vigor hypothesis and demonstrate why modeling insect detection is important in plant–insect studies. Weevils responded positively to water stress associated with seasonal dry‐downs, and this response was phenologically compartmentalized by larval feeding mode. Weevils preferentially attacked large and/or flowering stems, imposing an ecological cost on willow reproduction via increased stem mortality and susceptibility to future attack. We propose that the dual response to host plant stress and vigor is due to the synchronization between young weevil larval feeding and willow nutrient pulses that are mediated by environmental stress. In turn, this process drives successional dynamics, causing the juvenilization of upland willow plants and possibly delaying establishment of a willow‐dominated upland sere. These results highlight the common, but often overlooked, phenological basis of the plant stress and plant vigor hypotheses, which both focus on how stress changes the quality of plant resources available to immature insects.