Abstract

The plant apparency hypothesis predicts that apparent plants invest in broadly effective defences such as tannins while unapparent plants invest in specific toxins such as alkaloids. The stress hypothesis states that plants invest in cheaper defences if they have evolved in habitats that impose abiotic limitations to plant fitness. We tested these hypotheses by determining the concentrations of alkaloids and tannins in a representative sample of the vascular plants of continental Chile (with exclusion of Pteridophyta, Cactaceae, and Poaceae) consisting of 396 species. In a subsample of 166 species which contained both alkaloids and tannins, we constructed the A/T index (A/T = [alkaloids]/ [tannins]). We discarded the presumed effect of phylogeny (as estimated by taxonomy) on the variation observed in the data because no correlation of A/T with taxonomic relationships among species either at family or genus levels was found in a nested ANOVA with genera nested in families. Concentration of alkaloids was negatively correlated with that of tannins. We compared the value of A/T among species differing in life form (herbs, shrubs or trees), herb longevity (annual or perennial), leaf-shedding manner of woody plants (deciduous or evergreen), latitudinal range, and level of water stress typical in their natural habitat. Unapparent plants (herbs, annual) exhibited higher mean A/T index than apparent plants (shrubs and trees, perennial). A/T did not correlate with latitudinal range. Mean A/T values decreased from deserts to deciduous forests. The comparisons were not always significant due to the inevitable unbalance of the data set which lowers the power of the statistical tests employed. The results suggest that chemical defences are indeed distributed in a non-random manner among plants, and that to a large extent the predictions derived from the apparency and stress hypotheses are sustained. © Birkhäuser Verlag, 2006.