It is one thing to show that herbivorous insects affect plant performance. It is an entirely different matter to demonstrate that insect herbivory affects plant population dynamics. There is a vast literature on insect pests of crop plants that shows how attack by defoliating, sucking, stem-mining, and gall-forming species can delay seed ripening, reduce seed production and individual seed weights, reduce the rates of shoot and root growth, increase the susceptibility of plants to disease, and reduce the competitive ability of plants relative to their un attacked neighbors. This literature tells us virtually nothing, however, about the importance of insects in natural communities, chiefly because we have so little information on the regulation of plant populations in the wild. For example, we do not know whether plant recruitment is seed limited , so we cannot predict whether there would be an increase in plant numbers if the simple experiment of sowing extra seeds were carried out. Information on the nature of population regulation is vital because if plant recruitment is not seed limited, then insects that reduce seed production will not have an important effect on plant population dynamics.This review concentrates on material published from 1984 to 1987. Earlier material has been covered in other reviews (36, 64, 70, 94, 107, 12 1, 182, 206) . The first part examines the impact of insect feeding on different aspects of plant performance. The second part examines two of the best sources of evidence on the impact of insect feeding on plant population dynamics: (a) the release of specialist insect herbivores against target weed species in classical 531 Further ANNUAL REVIEWS 532 CRAWLEY biological control projects and (b) the exclusion of insect herbivores from natural plant communities by the repeated application of chemical in secticides. The limitations and shortcomings of these bodies of evidence are also discussed.
BackgroundUntil recently, conventional wisdom suggested that because the world was green, it was not possible that insect herbivores could be food limited (82, 167) . The dramatic outbreaks of herbivorous insects that sometimes followed the indiscriminate use of broad-spectrum insecticides (51) lent credibility to the view that populations of insect herbivores are regulated at low densities by the actions of various natural enemies, notably predators, parasitoids, and diseases (87). At these low densities, the argument went, it was unlikely that insects could have an important effect on plant popUlation dynamics.The flaws in these arguments are numerous and have been discussed elsewhere (36, 62, 162). They can be summarized in three themes: (a) The world is not always green. (b) All that is green is not edible. (c) What is edible is not necessarily of sufficiently high quality to allow increase of the herbivore population. The evidence from natural populations does not provide convinc ing evidence in favor of the widespread importance of regulation by natural enemies (107) except in a handful ...