termed ‘‘keystone,’’ eventually undermined the usefulness of a potentially powerful concept (Mills et al., 1993) Here, these issues are evaluated and a scheme of classification and definition (slightly modified from that offered by Power et al., 1996) is suggested, which further clarifies and advances the usefulness and application of the concept of keystone species Reevaluating the Keystone Species Concept The seeming lack of a working definition of keystone species prompted Mills et al (1993) to argue that careless usage had sufficiently degraded the value of the concept to justify its removal from usage They further argued that because the term was poorly defined it was functionally useless and conservation strategies therefore should not be based on protecting keystone species They advocated focusing on interaction strength rather than on a species’ keystone or nonkeystone status as a more useful management strategy In response to the criticism by Mills et al (1993), a group of ecologists with expertise in the study of keystone species and strongly interacting species met to evaluate the keystone species concept The publication that resulted from this meeting was a signal achievement, and much of the present article is patterned after this synthesis (Power et al., 1996) The group agreed with Mills et al (1993) that through misapplication and questionable redefinition, ecologists and conservation biologists had obscured the meaning of the term keystone species Rather than abandoning the concept, however, the group proposed clarification and adherence to a set of more sharply defined concepts for community dynamics They defined a keystone species as ‘‘one whose impact on its community or ecosystem is large, and disproportionately large relative to its abundance’’ (Power et al., 1996) This definition retained the essence of Paine’s original usage, but expanded it more broadly to include species other than predators Power et al (1996) stressed the importance of having a rigorous, quantitative method of assessing the community- or ecosystem-level effects of a species when determining if it is a keystone To assess a species impact on a community, they proposed a community importance (CI) index, which in practical terms is quantified experimentally as CIi ẳ ẵtN tD ị=tN 1=pi ị where pi is the proportional abundance of the species i before it was deleted, tN (for ‘‘normal’’) is a quantitative measure of a community or ecosystem trait under usual conditions (e.g., productivity, nutrient cycling, species richness, and relative abundances of species), and tD (for ‘‘deleted’’) is the trait in the absence of species i Keystone species are those whose CI is large relative to that of other species (Figure 6) On the basis of these considerations, Power et al suggested that if the total impact of a species is plotted against its proportional abundance, keystone species would cluster toward the upper left of the graph in Figure Power et al (1996) defined another group of species, ‘‘dominants,’’ as those that had large total impacts but in proportion to their abundance – these would lie to the upper right of the abundance–impact diagram Importantly, keystone species are not necessarily simply ‘‘strong’’ interactors, at least as the terms are defined here Frequency Keystone Species (a) Frequency 448 − (b) + Community importance Figure Possible frequency distributions of community importance (CI) values for all species in a community Positive values occur when a community trait decreases after a species is deleted For instance, in the absence of a mutualist, the target dominant species would also decrease Negative values occur when a community trait increases after removal of a species For instance, in the absence of a consumer, the target dominant species would increase Community importance values may be normally distributed around zero (a), indicating that most species have small effects and keystones are rare In some communities (b), CI values may have several modes, with keystone species indicated by values far from zero Reproduced from Power ME, Tilman D, Estes JA, et al (1996) Challenges in the quest for keystones Bioscience 46: 609–620 Keystone species not only have disproportionately large effects, but also have a community or ecosystem-wide impact through direct and indirect effects cascading through the system (Power et al., 1996; see Figure 1) Strong interactors are species that have a large impact on the species with which they interact, but this large impact could affect only a single interacting species Only when strong interactors have multispecies effects that alter the structure of the community, or the functioning of the ecosystem, are they also keystone species Related Concepts The idea that not all species have equal significance in community dynamics is an old one, and many efforts have been made to assign names or terms to distinguish species or groups of species with important roles Elton (1927) advanced the concept of ‘‘key-industry species’’ as single species of animals supporting a large number of consumers (e.g., copepods,