John N. Thompson. 2005. University of Chicago Press, Chicago. × + 400 pp., 11 halftones, 94 line drawings, 7 tables. ISBN 0-226-79761-9 (cloth) and ISBN 0-226-79762-7 (paper). Cloth, $75.00; paper, $28.00.—I have occasionally received comments from reviewers of my work on hummingbirds and their food plants to the effect that specialization and coevolution are the exception rather than the rule in pollination systems or, alternatively, that coevolution is such a diffuse process that it cannot be analyzed. These comments reflected the reviewers' frustrations from their studies of coevolution at the scale of local populations, which often failed to turn up evidence for reciprocal selection. During the past decade, however, the view of coevolution as either a rare or diffuse process has changed, largely through the influence of John N. Thompson and his theory that coevolutionary processes are best understood by studying them on a geographic scale. In his latest book, The Geographic Mosaic of Coevolution, Thompson reviews his theory and the recent evidence in support of it, and discusses its implications for coevolutionary interactions.

Part 1 addresses the framework of the geographic mosaic theory of coevolution (GMTC), beginning with an outline of Thompson's thesis. The next four chapters examine evidence for the major assumptions of his theory, especially differences between populations in genetic structure and outcomes of ecological interactions. In the next two chapters, Thompson presents the theory's assumptions, hypotheses, and predictions, and applies them to understand diversifying coevolution and speciation. Part 1 concludes with a chapter on how the GMTC can be analyzed, in which Thompson discusses 11 forms of evidence for coevolution. Students will want to have a copy of this list taped to their computer monitors as a guide.

Part 2 examines specific hypotheses on coevolution. Three chapters discuss antagonistic interactions and how the geographic mosaic maintains genetic polymorphisms, produces multispecies networks of antagonistic trophic interactions, molds levels of virulence and resistance, and contributes to the dynamics of sexual reproduction. The next three chapters consider how the geographic mosaic may lead to convergence of traits in mutualistic interactions and networks. One chapter is devoted to coevolutionary character displacement. Thompson concludes with a discussion of applied coevolutionary biology. (The book also includes an appendix summarizing all the major hypotheses and predictions.)

Thompson's GMTC makes three hypotheses about coevolution. First, natural selection differs among populations because of differences in how the fitness of one species depends on the other. Second, reciprocal selection occurs only within some local communities (“coevolutionary hotspots”) embedded within a matrix of coldspots where local selection is nonreciprocal, creating selection mosaics. Third, because of trait-remixing, species-level coevolved traits will be few in number.

At the simplest level, the GMTC is an argument about the scale of coevolutionary processes. Can the evolution of mutualisms and antagonisms be explained entirely from studies of single populations, or are studies at a larger scale necessary? But the GMTC also emphasizes the complex nature of the dynamics of coevolutionary interactions. Because populations differ in genetic structure and the outcome of ecological interactions, studies of single populations of interacting species may fail to capture the scope of the coevolutionary interaction and the intermixing of traits that results from population structure. Thompson's GMTC thus provides a template for how researchers should design their studies. Rather than begin studies of the coevolutionary process through detailed microevolutionary analyses of interactions in a single population, researchers should first conduct a geographic survey of the interaction and then focus subsequent efforts on areas with and without reciprocal interactions (coevolutionary hotspots and coldspots) and on the trait-remixing that may occur between them. A geographic approach reveals how a third party mediates diversifying coevolution between Red Crossbills (Loxia curvirostra) and lodgepole pine (Pinus contorta; Benkman 1999), and how specialization by sexes of hummingbirds changes with the relative frequency of their food plants (Temeles and Kress 2003).

Given his efforts to explain coevolution resulting from all forms of ecological interactions at all scales of organization, there are bound to be some weak spots in the book, and Thompson is at his weakest when he lacks sufficient data to support or interpret his hypotheses. For example, in Chapter 11, he presents his hypothesis of coevolutionary alternation, which maintains that predators preferentially attack prey species with low levels of defense, leading to selection for increased levels of defense in such prey. Because defense costs impose fitness costs in the absence of predation, selection will favor reduced defenses in prey species that are not currently attacked. As the relative levels of defense among prey species change over time, predators attack prey that are currently undefended. Some support for this hypothesis is provided by Davies and Brooke's (1989 a, b) studies of host alternation by Common Cuckoos (Cuculus canorus). Still, I wondered how this hypothesis might apply more generally to interactions between predators and prey, as well as how it might be reconciled with decisions at the behavioral level (e.g. optimal foraging theory). To his credit, Thompson notes that he may be wrong, and states that his main purpose is to push discussion of these interactions beyond that of “diffuse” coevolution, which provides no hypothesis for how these multispecies coevolutionary interactions actually evolve. I applaud him for trying, though in these discussions he strays from his geographic message.

The book is at its strongest where Thompson has evidence to support his theory. Thompson notes that he favored the inclusion of studies that have appeared in the 11 years since he published his second book, The Coevolutionary Process (Thompson 1994). Even readers familiar with Thompson's work will value his interpretation of recent coevolutionary studies by Benkman, Lively, Brodie, Burdon, Thrall, and others, and what is especially impressive is how masterfully Thompson moves from discussions of gene-for-gene matching and the evolution of sex in parasite-host systems to community analyses of plants and their pollinators. These discussions make good meat for graduate student seminars and provide a ton of material for faculty lectures.

All writers know how difficult it is to begin and end an article, and Thompson succeeds with both. In Chapter 1, he discusses the pervasiveness of coevolutionary interactions on Earth, from coevolved symbiotic interactions that led to mitochondria and chloroplasts, to primary succession that relies on nitrogen-fixation symbioses between rhizobial bacteria and legumes, to feeding on plant tissue made possible for many vertebrates and invertebrates by obligate coevolved symbionts in their digestive tracts, and about six or seven others. These examples of coevolutionary processes in the world around us are how Thompson introduces the subject in public lectures, and they make a powerful statement for why everyone should appreciate evolution. Readers looking for a way to introduce evolutionary biology to naïve students in an introductory biology course have much to gain from this chapter.

The last chapter, on applied coevolutionary biology, is similarly inspirational. As humans continuously move genes and species around the world, we disrupt existing coevolutionary dynamics and create new ones, especially at the geographic level. Similarly, by fragmenting habitats, we may shift coevolutionary dynamics from a geographic to a local scale in the absence of gene flow. Thus, humans are creating new systems for tests of the GMTC.

To a great extent, current evidence in support of Thompson's theory comes from studies of two-species interactions, with occasionally a third party. Looking at how such simple interactions form multispecies networks will be the next critical step in understanding the coevolutionary process.

In sum, given its attempt to understand coevolution at all levels, from viruses and hosts to lions and their prey, The Geographic Mosaic of Coevolution belongs on everyone's bookshelf. Readers will not be disappointed.