WITHIN-TREE AND AMONG-TREE VARIATION IN LEAF CHARACTERISTICS OF MOUNTAIN BIRCH AND ITS IMPLICATIONS FOR HERBIVORY

We employed a nested sampling design to quantify hierarchical patterns of variation in leaf characteristics of the polycormic tree, Betula p ubescens ssp. tortuosa. Sampling levels included trees, ramets within trees, branches within ramets, short shoots within branches, and leave s within short shoots. Leaf water content, specific weight, toughness, and nitrogen content were measured. We used only short shoot leaves, i.e. leaves of the same age. All hierarchical levels contributed signi ficant components of variation. Trees, and ramets within trees, usuall y contributed more variation than branches or shoots. Trees accounted for 12 to 64% of the total variance (least in toughness, most in nitro gen content). Within-tree variation due to different hierarchical leve ls was surprisingly large: 40 to 44% of the total variance in water co ntent, specific weight, and toughness was explained by ramets, branche s, and shoots. For water content and toughness, ramets within trees we re actually more variable than trees. Variance estimates were quite co nstant between years. Extensive within-tree variation apparently arise s from the organization of trees into modules that are more or less ph ysiologically autonomous and may even differ genetically due to somati c mutations. Unlike other traits. leaf nitrogen was invariant across r amets, indicating that processes of nitrogen allocation tend to be rel atively well integrated on a whole plant basis. Complex patterns of in tra-specific variation in their host plants complicate the oviposition and foraging behavior of herbivores. Results suggest the need for stu dies that treat variance as a quantitative attribute of individual tre es. We offer some recommendations for optimizing sampling efficiency, and avoiding pseudoreplication, in experiments that involve trees.