Alternative approaches to the analysis of comparative data: compare and contrast

Abstract

Several goals of evolutionary biologists can be approached by comparisons of species that occupy different habitats or that have evolved distinct suites of life history characters. One is to determine the ecological conditions favoring particular morphological or life history traits, and another is to detect adaptive combinations of traits that have ‘‘coevolved.’’ Recently, due to the development of statistical tools that use hypothesized phylogenetic relationships to detect the joint evolution of life history and ecological attributes among taxa, the comparative analysis of plant life history traits has enjoyed increased attention. Silvertown, Franco, and Harper responded to this interest by convening in 1996 a Royal Society Symposium aimed in part to highlight alternative approaches to the analysis of comparative data. This book represents the outcome of this meeting and includes 15 chapters, most of which use comparative data to interpret patterns involving the distribution of plant life history, mating system, or reproductive traits across taxa and environmental conditions. In addition, one chapter (by Michael Donoghue and David Ackerly) explores the effects of uncertainty in phylogenetic reconstruction on the outcome of one statistical test that makes use of phylogenetic information. Three chapters provide reviews of mathematical models: one concerning the allocation and packaging of resources among flowers and fruits (by Lawrence Venable); a second evaluating the relative importance of the automatic selfing vs. reproductive assurance hypotheses for the evolution of self-pollination and the use of coalescent theory to detect evolutionary events associated with the evolution of self-fertilization from outcrossing ancestral taxa (by Daniel Schoen, Martin Morgan, and Thomas Bataillon); and the third exploring the evolution of life history traits in heterogeneous environments (by Richard Sibly). It is tough to identify the best audience for this book. The chapters are neither review articles nor rigorous primary research articles. Most summarize new data analyses, but in few cases do they provide sufficient statistical and methodological detail to allow the reader to evaluate fully their appropriateness or interpretation. Consequently, I would not recommend this book as the main focus, say, of a graduate seminar, unless it were to be accom