Most contemporary biologists think of species as evolutionary units held together by gene flow. For instance Mayr (1) writes" The nonarbitrariness of the biological species is the result of... internal cohesion of the gene pool." Merrell (2) states" The species is a nat-ural biological unit tied together by bonds of mating and sharing a common gene pool." This idea is founded in the pioneering work of Dobzhansky, Mayr, Stebbins, and others integrating the theory of population genetics with laboratory and field experiments and observations to produce the neo-Darwinian or synthetic theory of evolution. These workers quite logically concluded that differentiation of populations would be prevented by gene flow, and they focused their discussions of speciation on various means of interrupting that flow. In other words, they empha-sized the role of mechanismsisolating populations fromone another. Until quite recently there has been little rea-son to question thisview. In the past few years, however, growing evidence from field experiments has led us to reevaluate the processes leading to or-ganic diversity, and to conclude that a revision of this section of evolutionary theory is in order.

In this paperwe suggest that many, if not most, species are not evolutionary units, except in the sense that they (like genera, families, and so forth) are prod-ucts of evolution. We will argue that selection is both the primary cohesive and disruptive force in evolution, and that the selective regime itself determines what influence gene flow (or iso-lation) will have. Threefold evidence is presented for this. We will show that (i) gene flow innature is much more restricted than commonly thought;(ii) populations that have been completely