In his recent review of the developing theory of insect-plant interactions, Gilbert (52) identified four major thrusts in research: insect-plant coevolution, host plants as islands, plant apparency and chemical defense, and resource predictability versus evolutionary strategies of insects. From his review it is evident that, with a few exceptions (23, 45, 53, 64), devdoping theory is addressing primarily a two trophie levd system. In reality, of course, all terrestrial communities based on living plants are composed of at least three interacting trophic levels: plants, herbivores, and natural enemies of herbivores. We argue that theory on insect-plant interactions cannot progress realistically without consideration of the third trophic level. A closer look at the mechanisms of interactions reveals a paradox, and plants have many effects, direct and indirect, positive and negative, not only on herbivores but also on the enemies of herbivores. The third trophic level must be considered as part of a plant's battery of defenses against herbivores.

No abstract

Five hypotheses were invoked to account for variation in galling species number per location on plants of different structural complexity, namely herbs, shrubs, and trees, both in Brazil and USA. The hypotheses were: 1) the altitudinal/latitudinal gradient hypothesis; 2) the harsh environment hypothesis; 3) the plant species richness hypothesis; 4) the host plant area hypothesis; 5) the plant structural complexity hypothesis. The altitudinal and the harsh environment hypotheses were correlated and supported with sample data in both localities, with increasing gall species number as altitude/latitude declined and as sites became hotter and drier. The two hypotheses were separated by studying riparian sites and dry hillside sites at the same elevation in Arizona. Galling species frequency was higher in dry sites than in riparian sites, supporting the harsh environment hypothesis. Of the five hypotheses tested only the harsh environment hypothesis predicted that galling insect species number should vary in response to environmental variables such as moisture and temperature. Temperate shrubs supported more galling species than did other plant types, both in dry and mesic sites. The overall difference between galling species richness for tropical and temperate latitudes was not statistically significant. Free-feeding insect herbivore species exhibited the opposite pattern of species richness to gallers, being more speciose in riparian sites. The present study corroborates the hypothesis that the gall forming habit is an adaptation to harsh or stressful environments, and we describe for the first time broad scale geographical patterns in galling insect species richness.

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';

Abstract. We evaluate a three‐part hypothesis explaining why gall‐inducing insect species richness is so high in scleromorphic vegetation: (1) persistence of low nutrient status scleromorphic leaves facilitates the galling habit in warm temperate latitudes; (2) favourable colonization sites for gallers result from reduced hygrothermal stress, high phenolics in the outer cortex of the gall, and reduced carnivore and fungal attack in the gall; and (3) in more mesic sites, mortality is high due to carnivore attack and invasion of galls by fungi. Over 280 samples of local species of galling herbivorous insects from fourteen countries on all continents except Antarctica revealed a strong pattern of highest richness in warm temperate latitudes, or their altitudinal equivalents. The peak of galling species richness on the latitudinal gradient from the equator into the Arctic was between 25 to 38° N or S. Galling species were particularly diverse in sclerophyllous vegetation, which commonly had greater than twelve species per local sample. In mesic, non‐sclerophyllous vegetation types the number of galling species was lower with twelve or fewer species present. Many sites in sclerophyllous vegetation supported between thirteen and forty‐six galling species locally, including campina islands in Amazonia, cerrado savanna in central Brazil, the Sonoran Desert in Arizona and Mexico, shrubland in Israel, fynbos in South Africa and coastal scleromorphic vegetation in Australia. At the same latitude, or its elevational equivalent, galling species richness was significantly higher in relatively xeric sites when compared to riparian or otherwise mesic habitats, even when scleromorphic vegetation dominated the mesic sites. The results were consistent with the hypothesis and extend to a more general level the patterns and predictions on the biogeography of gall‐inducing insects.

No abstract

We studied the relationship between habitat moisture and gall-forming insect populations. Population sizes for most galling taxa were significantly larger in xeric habitats compared with mesic habitats. Our results indicate that the differential abundance of galling insects in these habitats is due primarily to differential mortality and survivorship. Mortality factors acting upon eight insect galling species (belonging to eight genera and four families) were measured on six species (five genera and five families) of host plants. Survival was significantly higher for galling populations inhabiting xeric habitats compared with mesic habitats. Parasitism was higher in mesic habitats in seven of eight habitats and fungus-induced diseases were higher in five of seven habitats. Mortality due to predation and other (unknown) factors showed no clear trends. Overall, there was a tendency towards lower mortality and consequently higher survival for populations inhabiting xeric habitats. We hypothesize that reduced mortality caused by natural enemies and endophytic fungi has contributed to the speciation and radiation of galling insects in apparently harsh environments.

To investigate the relationship between oviposition preference and offspring performance we tested three hypotheses about the oviposition behavior of Euura lasiolepis: (I) the preference hierarchy hypothesis that Euura had an oviposition preference based on shoot lengths; (2) the larval survival hypothesis that the preference hierarchy corresponded to the suitability of shoots for larval survival; and (3) the flexibility hypothesis that the preference hierarchy changed as resource availability changed. The oviposition preference hierarchy of Euura lasiolepis, a shoot-galling sawfly, for plant shoots of different lengths strongly corresponded with the rank of shoot lengths for larval survival. The oviposition preference hierarchy was related to shoot growth; the more rapidly a shoot was growing the higher the probability that an oviposition site on a shoot would be attacked. Larval survival was highest on long shoots on young ramets. The probability of shoot abscission was negatively related to shoot length, and Euura in galls on abscised shoots invariably died. Intraspecific interactions and willow clone identity, independent of shoot length and ramet age, had relatively small influences on larval survival. The oviposition hierarchy was flexible; iflong shoots were not available shorter shoots were accepted for oviposition. Shoots were always attacked in order of length, so that a strong relationship between preference and performance was always maintained.