The functioning of ecosystems can be strongly driven by landscape attributes. Despite its importance, however, our understanding of how landscape influences ecosystem function derives mostly from species richness and abundance patterns, with few studies assessing how these relate to actual functional rates. We examined the influence of landscape attributes on the rates of herbivory in seagrass meadows, where herbivory has been identified as a key process structuring these relatively simple systems. The study was conducted in three representative Posidonia oceanica meadows. The principal herbivores in these meadows are the fish Sarpa salpa and the sea urchin Paracentrotus lividus, and we hypothesized that differences in their interaction with landscape attributes would significantly influence herbivory rates. We measured herbivore abundance, herbivory rates, primary production and plant quality (C:N) in seagrass patches embedded either in rock or in sand (matrix attribute), in patches either near or far from a rocky reef (distance attribute) and at the edges and interior of meadows. Our results show that matrix and meadow edges significantly affected the actual levels of herbivory. Herbivory rates were higher in seagrass patches embedded in a rocky matrix compared to those on sand, and herbivory at the centre of seagrass meadows was higher than at the edges. In contrast, patch distance to rocky reefs did not affect herbivory. Neither herbivore abundance nor food quality explained the patterns across different landscape attributes. This suggests that variation in herbivory across the landscape may be related much more to behavioural differences between species in their evaluation of risk, movement and food preference in relation to the landscape structure. Our results indicate that richness and abundance patterns may mask critical interactions between landscape attributes and species responses, which result in considerable heterogeneity in the way key functional processes like herbivory are distributed across the ecosystem mosaic.