Leaves or their functional analogues provide outstanding opportunities for comparative studies. Here, I use leaves to illustrate the crucial role of ecological, biogeographic and phylogenetic comparisons in generating and testing hypotheses regarding the adaptive significance of morphological variation, the relative importance of selective pressures vs phylogenetic constraints, and the rise of adaptations within lineages. The complementary roles of comparative studies and optimality models are stressed throughout.

The first section of this paper reviews 23 ecological patterns in leaf form, physiology and arrangement which have been uncovered by comparative studies. Three general sets of energetic trade-offs, involving the economics of gas exchange, support, and biotic interactions, appear likely to influence the evolution of leaves and underlie these trends. The first of these trade-offs is illustrated with an analysis of the adaptive significance of leaf size, in both terrestrial and aquatic plants. The resulting predictions are compared with the actual trends observed, and the relative strengths and limitations of the approach are discussed.

The second section addresses the role of selective pressures and phylogenetic constraints in determining features of leaf form and phenology in forest herbs. Ecological comparisons of 74 species from a site in the Virginia Piedmont show that members of each temporal photosynthetic guild display evolutionary convergence in several aspects of leaf form and arrangement. These convergences can each be understood in terms of models that assume that selection favours plants whose form and physiology tend to maximize whole-plant growth. Phylogenetic comparisons indicate that congeners of guild members share the same leaf phenology as the guild members themselves. This remarkable finding suggests that phenology is evolutionary rather non-labile within genera but that, within guilds, species in several different genera and families converge strongly in other leaf traits. In this case, phylogenetic constraints appear to be important mainly in determining which lineages evolve particular phenologies and leaf adaptations, not whether they arise. The section concludes with a critical discussion of the capacity of two classes of evolutionary models, based primarily on functional considerations or phylogenetic constraints, to produce truly deductive predictions.

The third section briefly reviews an analysis of adaptive radiation in leaf shape among violets of eastern North America. Each ecological group of species displays the leaf shape expected on functional grounds and separate lineages (as judged by means of phylogeny independent of leaf shape) show parallel trends. Ecological and phylogenetic comparisons, in combination with optimality models, provide insights into the sequence of habitats invaded and leaf forms evolved.

I conclude with comments on the advantages and limitations of comparative studies, and speculate on avenues for future research on leaf form. An integrated approach, involving comparisons at ecological, phylogenetic and biogeographic levels, and complemented by optimality analyses and detailed populational studies involving biomechanics, physiological ecology and ecological genetics, is strongly advocated.