Premise of research. Seed production by flowering plants depends on abiotic and biotic factors whose interacting effects may be hidden. We previously reported that exposure to dust from unpaved roads reduced the average amount of pollen on flowers of Ipomopsis aggregata but did not consistently reduce mean seed set per fruit. Here we explore one possible explanation—that the expected detrimental effect of lower pollen loads on seed production was obscured because dust reduces not only pollination success but also attack by a predispersal seed predator. Methodology. Over three consecutive summers, we scored the fates of more than 4000 flowers, comparing those on hand-dusted I. aggregata plants with those on clean control plants in the same natural populations. We censused plants throughout each flowering season, marking phenological cohorts of flowers and examining them for eggs of the anthomyiid fly Hylemya sp. We subsequently scored expanded fruits for seed loss to Hylemya larvae. Pivotal results. Control plants consistently had higher egg loads and rates of seed loss to larvae than dusty plants. Furthermore, most eggs appeared early in the season, when flowers are most likely to set fruit and ovule numbers are highest. In agreement with earlier studies, unparasitized fruits from dusty plants did not contain fewer seeds than fruits from control plants. But when we simulated a system without Hylemya by assigning each parasitized fruit the average number of seeds for unparasitized flowers in the same phenological cohort, control plants outproduced dusty plants. Conclusions. Predispersal seed predation likely contributed to the previous lack of a negative effect of dust on seed set in I. aggregata. This study illustrates how exploring unexpected experimental outcomes can reveal important ecological interactions and how the mechanistic understanding so gained can improve the ability to generalize results to other systems.