Mimulus luteus (Scrophulariaceae) is a perennial herb occurring in the South American Andes that shows a wide variation in the size and shape of a red spot on the lower lobe of the yellow flower. We describe the preference of four insects (three bees and one butterfly) and one hummingbird species for floral characters, and estimated the strength, direction, and form of pollinator‐mediated selection through female fitness. We applied geometric morphometrics to describe the preference of pollinator species for different guide shapes. Our results revealed striking differences in the floral phenotypes preferred by insects and hummingbirds. Insects visited flowers with corollas 1.25‐fold larger and guides 1.72‐fold larger than the hummingbird species did. While insects preferred flowers with nectar guides pointing toward the corolla tube, the hummingbird preferred flowers with heart‐shaped nectar guides. Most of the floral preferences shown by pollinators translated into significant linear and nonlinear selection coefficients. When selection was analyzed on a per‐flower basis and for female fitness, corolla size was under positive directional selection, and nectar guide size and shape were under disruptive selection. Because the insect and hummingbird pollinators showed a strong segregation in their daily activity time, we suggest that current disruptive selection on the nectar guide phenotype can result from the differential availability of the rewarding floral variants over a day. Our findings suggest that pollinator‐mediated selection favoring extreme phenotypes in M. luteus may not only contribute to high nectar guide variation found in this species, but also can promote divergence of corolla and nectar guide traits.

Understanding the factors affecting parasite aggregation in natural host populations is one of the central questions in parasite ecology. While different biological mechanisms giving rise to aggregation have been documented in the literature, the role of established parasites in vector attraction, and its importance in determining clumped parasite distributions has received less attention. In a two-year field study, we evaluated the importance of a bird vector, Mimus thenca (Mimidae), on the aggregation dynamics of the holoparasitic mistletoe, Tristerix aphyllus, on its cactus host, Echinopsis chilensis. Removal of T. aphyllus from cacti decreased the number of visits and the time spent by the bird vector, which resulted in a 3.5-fold lower seed deposition of the mistletoe on experimental hosts than on control hosts. Vector preference, however, was not the only factor affecting aggregation in this system. Spine length of the cactus acted as a first line of defense against parasitism, by discouraging bird perching on top of host columns. While heavily parasitized hosts received more seeds than unparasitized hosts, spines counteracted this effect. These results provide field evidence that parasite aggregation results from the balance between vector behavior and host resistance traits.