Species in biology are traditionally perceived as kinds of organisms about which explanatory and predictive generalizations can be made, and biologists commonly use species in this manner. This perception of species is, however, in stark contrast with the currently accepted view that species are not kinds or classes at all, but individuals. In this paper I investigate the conditions under which the two views of species might be held simultaneously. Specifically, I ask whether upon acceptance of an ontology of species as diachronic segments of the tree of life (this is one version of the species as individuals ontology) species can perform the epistemic role of kinds of organisms to which explanatory and predictive generalizations apply. I show that, for species-level segments of the tree of life, several requirements have to be met before the performance of this epistemic role is possible, and I argue that these requirements can be met by defining species according to the Composite Species Concept proposed by Kornet and McAllister in the 1990s.