RESUMO
Populations of the same species vary in their secondary metabolite content. This variation has been attributed to biotic and abiotic environmental conditions as well as to historical factors. Some studies have focused on the geographic variation of chemical diversity in plant populations, but whether this structure conforms to a central-marginal model or a mosaic pattern remains unclear. Furthermore, assessing the chemical diversity of invasive plants in their native distribution facilitates the understanding of their relationships with natural enemies. We examined the geographic variation of chemical diversity in Mexican populations of the bittervine weed Mikania micrantha and its relationship to herbivore damage. The foliar volatile terpenoid blend was analyzed in 165 individuals of 14 populations in the Pacific and Gulf of Mexico tropical watersheds. A cluster analysis grouped individuals with similar terpenoid blends into 56 compositional types. Chemical diversity was measured using the number of compounds and their concentration within the blends for individuals, and the number and frequency of compositional types for populations. A stepwise multiple regression analysis performed with geographic, climatic, and chemical diversity variables explained herbivore damage. However, population-level chemical diversity was the only variable found to be significant (ß = -0.79, P = 0.042) in the model (R(2) = 0.89). A Mantel test using Euclidean distances did not indicate any separation by geographic origin; however, four barriers were identified using Monmonier's algorithm. We conclude that variation in population-level chemical diversity follows a mosaic pattern in which geographic factors (i.e., natural barriers) have some effect and that variation is also associated with the local intensity of herbivore attack.