GRADIENT ANALYSIS USING PLANT MODULAR STRUCTURE - PATTERN IN PLANT ARCHITECTURE AND INSECT HERBIVORE UTILIZATION

Distribution of five species of tropical insect herbivores was recorde d in relation to gradients of modular length on their host plants. Pat terns in plant architecture and herbivore responses were compared with a well-studied temperate herbivore, Euura lasiolepis Smith (Hymenopte ra: Tenthredinidae). The five tropical species were each from a differ ent family, and host plants belonged to different families and vegetat ion types. Using modular gradients of shoot or leaf length we showed t hat plant architecture was comparable directly and was similar even wh en trees, shrubs, and grasses were compared. Also, the membracid, cica dellids, psyllid, and cecidomyiid in the tropics showed patterns of at tack on modular gradients consistent with the patterns derived from hi gh ovipositional preference linkage with larval performance, known to exist in the temperate species. Strong positive relationships between module length and percentage of modules attacked in each length class were observed, with variance accounted for ranging from 68 to 96%. One pest species on forage grasses in the on vicinity of Brasilia, Deois flavopicta Stal (Homoptera: Cercopidae), showed a weak response to the shoot-length gradient of its host grass (r(2) = 0.41) and an insignif icant response per unit length of shoot, suggesting a generalized capa city to use wide variation in module size. A method was developed for direct comparison of species employing a comparison between a hypothet ical ideal response to modular gradients and the actual response. The actual slope of the attack versus module length regression, expressed as a percentage of the hypothetical slope, provided estimates of herbi vore responses to modular plant structure consistent with hypotheses o n plant vigor and phylogenetic constraints.