Investigations on plant–animal interactions have traditionally focused on single interactions at a time (e.g. herbivory, pollination), yet plant fitness is generally influenced in complex ways by several interactions operating concurrently, and very little is known on the degree of spatial consistency of the direct and indirect effects that link different interactions. This paper evaluates experimentally whether direct and indirect effects of herbivory on male and female flower size and pollination success of the monoecious tropical shrub Cnidoscolus aconitifolius (Euphorbiaceae) remain consistent at three distant regions in Yucatan (southeastern Mexico). Plants were subjected to different levels of leaf defoliation, and treatment effects on floral traits (corolla area, corolla tube length, pollen production), and male and female components of pollination success (percent pollen removal, number of pollen tubes) were subsequently measured to evaluate the indirect effect of herbivory on plant reproductive success via pollination. Defoliation had significant direct effects on floral traits, but its indirect effects differed between the male and female components of pollination success. Moreover, the relationships between defoliation, floral traits and male and female pollination success varied spatially (i.e. between regions), although they were frequently more spatially-consistent for male success than for female success. Results from this study stress the importance of explicitly testing for spatial variation in direct and indirect effects arising from plant-animal interactions.