One of the most important factors affecting the fitness of insect herbivores is their diet that is, the quality of the plant species they eat. Polyphagous herbivores in particular face a challenge, as eating different host plant species can result in differences in life-history traits, such as growth, development time and fecundity. These differences may be due to them having a limited possibility to co-evolve with all of their potential host plants, which have differing chemical (nutritional value, secondary metabolites) and other traits (e.g. mechanical defence) that affect the herbivores’life-history traits (Gordon, 1961; Erickson and Feeny, 1974; Cates, 1980; Price et al., 1980; Berenbaum and Zangerl, 1999). Thus, it is likely that generalist herbivores are adapted to the most common secondary metabolites and are more sensitive to defensive compounds that only occur in some plant genera (e.g. Levins and McArthur, 1966). However, plant secondary metabolites are not always harmful to herbivores; some of them are used as feeding cues, especially by specialist herbivores, and some can be beneficial to the herbivore. Carotenoids, for example, are important antioxidants and reduce the harmful effects of stress caused by, for example, ultraviolet radiation or infection (Demming-Adams and Adams, 1996; Ouchane et al., 1997).

It has often been demonstrated that generalist herbivores perform differently on different host plant species (e.g. Price et al., 1980; Bernays and Chapman, 1994). In spite of the performance differences, genetic interactions in the performance of herbivores feeding on different host plants have generally not been found (see, for example, Jaenike, 1990 and references therein). Genetic interactions in herbivores’ growth or other general performance measures on different host plant species would suggest that there is a trade-off in the metabolism of allelochemicals between different host plant species. The absence of these interactions has been interpreted to mean that the ‘metabolic load’ of detoxifying capacity (which is expected to be energy limited) has by itself a trivial effect on larvae (Scriber and Feeny, 1979; Appel and Martin, 1992). Looking for energy costs is perhaps not the best way to seek to understand the feeding costs of herbivores on many different host plant species.