IntroductionSince Fraenkel (1959) proposed a leading role for plant secondary metabolites (PSMs) in the interactions between plants and herbivores, science has achieved broad insight into the diversity of PSMs and herbivores’ counter-adaptations to them (Freeland & Janzen, 1974; Foley et al., 1999; Foley & Moore, 2005). However, the more we learn about the distributions and functions of PSMs in natural systems, the sharper the limitations of our understanding become. In countless plant–herbivore interactions, ecologists have identified PSMs that act as feeding deterrents, toxins, digestibility reducers, feeding or oviposition cues, and signals for communicating to neighbouring plants and natural enemies of herbivores. However, most studies focus on the interaction between single species of herbivore and plant, usually with observations of captive animals fed diets containing PSMs under highly simplified conditions. Although such approaches are a necessary first step in isolating and characterising the actions of PSMs, they greatly oversimplify the complex interactions that occur between wild herbivores and plants. The next challenge for ecologists is to ‘scale up’ the roles of PSMs in plant–herbivore interactions, as we understand them from controlled experiments at small temporal scales, to predict ecological interactions at greater temporal and spatial extents.A captive herbivore may commonly eat less as PSM concentrations in its food increase, but can this predict the foraging decisions of a wild animal within its home range, or, ultimately the distributions and abundances of plant and herbivore species and genotypes? Scaling up has an obvious spatial component, because wild animals forage more extensively than do captive animals, but it also has a temporal component. Experiments usually describe plant–herbivore interactions over very short time intervals, but in nature they are continuous and the effects of PSMs can be long-lasting (Cheeke, 1998). Animal feeding preferences are dynamic and often change with season or reproductive state, or through the ongoing process of refinement of conditioned flavour aversions (Provenza, 1996). With increasing spatial extent and finer spatial grain size comes greater complexity in the interactions between plants and animals; PSMs are rarely distributed evenly throughout landscapes, and understanding how this influences plant–animal interactions requires approaches adopted from resource ecology, foraging theory and spatial ecology and often an extensive, high-resolution picture of the foodscapes within which animals forage (van Langevelde & Prins, 2008).