Plant Chemistry and Host Range in Insect Herbivores

Abstract

Bernays and Graham (1988) argue at the outset of their interesting paper that the role of plant chemistry in plant-herbivore coevolution has been overemphasized, and that natural enemies, especially predators, may be the dominant factor in the evolution of host range. There has, of course, never been any doubt that factors other than the biochemical characteristics of plants may help determine host range of herbivorous insects. Moreover, close relationships between insects and a narrow range of plants may be promoted by the evolution of concealment from predators in relation to a single background. The plant provides the substrate for the larvae, not just their food (Ehrlich and Raven 1964). Also well known is that even in cases of extreme specialist herbivores, nonchemical factors may finetune host ranges. For example, Holdren and Ehrlich (1982) concluded that the larvae of the butterfly Euphydryas editha in Colorado were restricted by ecological factors to Castilleja linariibfolia, even though two other nutritionally satisfactory, closely related host plants were available. All three plants species are members of the Scrophulariaceae, and presumably contain the iridoid glycosides prerequisite to any plant being a suitable Euphydryas host (Bowers 1983). We believe that the examples presented by Bernays and Graham do not yet allow us to reject the traditional understanding of the dominance of host biochemistry in the coevolution of plants and insect herbivores, and thus in determination of diet breadth. That arthropod herbivores can rapidly evolve new oviposition choices and host utilization abilities certainly does not contradict the idea that host plant chemistry plays a key role. After all, the ability of insect pests to evolve resistance to synthetic pesticides in 10 generations or fewer is legendary, and equivalent performances in response to plant chemicals are to be expected. Certainly, populations of some plants seem to be under selective pressure to produce an array of defenses that make it difficult for herbivores to evolve resistance (e.g., Dolinger etal. 1973). The observation that, in some cases, deterrent chemical constituents are comparatively harmless suggests that some herbivores are avoiding certain plants for reasons other than their toxicity. That, however, is not strong evidence that characteristics other than the edibility of plants (which has other components besides noxiousness) are mediating those plant-herbivore relationships. The preferential choice by grasshoppers of C3 plants over C4 plants, apparently because large fractions of the edible sheath cells of the latter cannot be digested (Caswell and Reed 1975, 1976), is but one example. Furthermore, evidence suggesting that grasshopper survival, size, and growth rate are greater on native vs. nonnative grasses implies a substantial degree of evolutionary response to the native grass defenses even among comparatively generalist insects (Joern 1988). As Bernays and Graham note, putative cases must be examined with great care. Nevertheless, suppose that a well-stratified sampling reveals that the deterrents that keep herbivores from eating harmless, edible host plants usually are nontoxic chemicals. Then we indeed will have to find new explanations (such as constraints applied by predators) for the restricted diets commonly observed (at least in butterflies, the most thoroughly studied major group of insect herbivores). We find that result unlikely, however. Moreover, examples accumulate of chemicals that make plants inedible to insect herbivores. For example, specialist insects that attack milkweeds (Asclepiadaceae) commonly cut the veins of the leaves they are eating, blocking the flow of latex to their feeding sites distad of the cuts (Dussourd and Eisner 1987). Experimentally cutting the veins makes milkweeds accessible to generalists that do not normally feed on them, while insects that eat a latex-rich leaf find their mandibles badly gummed. One must also be cautious in evaluating the selective impact of insects that are extremely in relation to their host plants. As Ehrlich and Birch (1967) pointed out, that impact is easily understated. For instance, the cactus Opuntia is no longer a plague in Queensland and the moth Cactoblactis, which was introduced to keep Opuntia under control, is extremely rare in relation to the cactus. The tiny lycaenid butterfly Glaucopsyche lygdamuus, for another example, might be thought scarce and inconspicuous in relation to its lupine host plants (overwhelmingly so if com-