Major hypotheses on the adaptive significance of insect gall formation are reviewed: nonadaptive, plant protection, mutual benefit, nutrition, microenvironment, and enemy hypotheses. We evaluate the validity of each, and find the first three to be without merit because galls clearly have adaptive features for the insect, but few if any for the plant, and the galler has negative impact on the plant, making the relationship parasitic. Predictions are developed to enable testing of hypotheses, and tests are discussed. Nutrition and microenvironment hypotheses are supported, while the enemy hypothesis remains with several uncertain issues to be resolved. The evolution of the galling habit has followed two pathways, one via mining plant tissues and the other from sedentary external herbivores that then modify plant growth. In each route the sequence of selective factors was probably different, but improved protection from hygrothermal stress and improved nutrition are of pr~mary importance, and protection from enemies probably reinforced the galling habit. KEY WORDS evolution. selection 'f';

No abstract

Two closely related species of ground crickets, Allonemobius fasciatus and A. socius, overlap and hybridize in a contact zone in the eastern United States. In earlier work, Howard (1986) described geographic variation in the width of the zone and in the strength of reproductive isolation between the two ground crickets. The zone was wider in the hills and mountains of southeastern Ohio and West Virginia than along the eastern coastal plain, and reproductive isolation appeared to be stronger where the zone was wider. Howard attributed the greater width in the mountains to the wide intermingling and patchy distribution of habitats appropriate for a species adapted to a northern climate and for a species adapted to a southern climate. He also pointed out that the mosaicism and the increased breadth of the zone in the mountains enhanced the probability ofoccurrence ofreinforcement. We tested three predictions that emerged from Howard's hypothesized links among topographic diversity, zone width, and the strength of reproductive isolation. The first two predictions were fulfilled. The northern cricket, A. fasciatus, occurred in the high mountains south of its previously known distributional limit; and the zone narrowed considerably in Illinois, an area oflow topographic diversity. These results provide further evidence for the importance ofthe environment in determining the structure ofthe zone. The third prediction was falsified. Contrary to the prediction, the strength of reproductive isolation between the two species was as strong in Illinois as in the Appalachian Mountains. This result suggests that if reinforcement has occurred in the zone, the width of the zone has not been a major factor in the process.

Populations of creosote bush (Larrea tridenrutu (DC)Coville), were studied in Arizona to determine whether associated gallformers (Cecidomyiidae: Asphondylia spp.) were more abundant on water-stressed or nonstressed plants. Gall densities were measured along a steep elevational gradient that extended from mesic, higher elevations to lower elevations in the desert; and in the Grand Canyon where severely water-stressed and relatively unstressed plants occurred adjacently. At the Grand Canyon site, the responses of creosote bush to water stress were also studied.2. The number and densities of Asphondylia species increased both at lower elevations and locally on water-stressed plants in the Grand Canyon, indicating that climatic and local conditions influence gallformer abundance in the same way.3. Five of the eight Asphondyfiu species studied at the Grand Canyon site were more abundant on stressed plants, two species were more abundant on nonstressed plants and one species showed no preference for either plant type.4. Densities of most species on stressed plants were positively correlated with the number of meristematic terminals per branch, which were more numerous on stressed plants, due to a bushier architecture. Flower gallformers were more abundant on nonstressed plants, which produced more flowers. Gall densities did not correlate with chemistry measurements, although these parameters also varied with level of stress. .These results suggest that gallforming species respond variably to plant stress, even within a closely-related lineage, and that there are effects of stress on plants, including architectural changes, that may be more important to herbivores than biochemical effects emphasized by White (1984) and others.

Predictions of the carbon-nutrient balance hypothesis were tested using a study of within-species phytochemical variation in the arroyo willow,Salix lasiolepis. The prediction that a balance between nutrients (total protein) and carbon-based secondary metabolites (total phenols) should exist was supported using water treatment and fertilizer experiments and wild willow clones. Leaf nitrogen content and net photosynthetic rates of plants potted in soil in which parental plants grew was low, indicating that wild plants exist under relatively low nutrient status-high carbon balance conditions. The hypothesis also correctly predicted positive relationships between shoot length and phenols in glasshouse plants, wild plants, and plants in the water treatment experiment and negative relationships between shoot length and phenols in the fertilizer treatment experiment. Total phenolic glycosides, fragilin, picein, salicortin, tremulacin, and tremuloidin all correlated positively with shoot length in glasshouse plants on a carbon-biased balance, and male willows had generally lower levels of phenolic glycosides than females. Salicortin and tremulacin showed the strongest positive relationships with shoot length.