Background/Question/Methods Plants produce a bewildering array of secondary metabolites, some of which are thought to mediate resistance to herbivores and pathogens, which may in turn result in an increased relative fitness for the plant. However, these defenses may incur significant fitness costs in terms of growth and reproduction. To cope with these potential costs, many defense-related compounds are only expressed in response to herbivory.  But if herbivore pressure is removed, the costs of constitutive defenses could outweigh benefits in terms of herbivore resistance.  Does natural selection by herbivory select for plants with increased constitutive defenses?  Will removal of natural selection by herbivory result in a plant community with increased growth rate and inducible defenses? To test the effects of natural selection on plant defenses, several plots of Solidago were sprayed repeatedly with insecticide for twelve years, while an equal number were left unsprayed.  Differences between constitutive and induced chemistry were compared, as well as growth rate in a common garden experiment. 
Results/Conclusions

Plants from under an herbivore exclusion regime have higher growth rates and lower constitutive defenses than plants under selection by herbivores.  These plants also exhibit induced resistance to Spodoptera exigua and inducible serine protease inhibitors, while plants under selection by herbivores do not. Herbivory selects for plants with higher constitutive defenses and without induced resistance to Spodoptera exigua.