Keystone Species kelp (Figure 4) During the 1990s, sea otter abundance declined sharply at several Aleutian Islands Once freed from sea otter predation, sea urchin populations increased, leading to severe declines in kelp and associated species Estes and coworkers suggested that killer whales may have shifted their diet from large marine mammals to sea otters ultimately as a consequence of human activity Overfishing and increased ocean temperatures (due to global climate change) are associated with declines in forage fish for sea lions and seals, and consequently these marine mammals have declined sharply in number Historically, these large marine mammals have been the primary prey of killer whales; in their absence, some orcas have shifted their diet to otters Because sea otters are small relative to pinnipeds, rates of predation loss have evidently been high, with substantial direct and indirect consequences for subtidal communities 30 Dipodomys removed Inside * * Dipodomys present 20 * Cover (%) 446 10 ** * 30 * ** * ** * * ** * * * ** Outside Generalization and Liberalization of the Definition: Other Types of Keystone Species 20 Cover (%) These earliest examples of keystone species were of consumers that modified the community through predation However, some investigators suggested that species other than top predators could play keystone roles in communities Subsequently, a variety of different types of organisms were termed ‘‘keystone’’ species, including herbivores, plants, pollinators, pathogens, habitat modifiers, and mutualists The remainder of this article presents examples of such alternative keystones, summarizes a critique of these ideas, and offers proposals for further clarification of keystone and other functionally important species 10 ERIN PERG ARAD ANNG BOAR BOBA Vegetation type Herbivores By consuming primary producers, keystone herbivores can have dramatic impacts on community structure Among species identified as keystone herbivores are kangaroo rats and bison In the Chihuahuan Desert, Brown (1998) (see summary) suggested that a guild of kangaroo rats played a keystone role In plots where kangaroo rats were excluded, grass cover increased threefold This was accompanied by a change in species composition; species shifted from those typical of desert shrubland to those characteristic of arid grasslands (Figure 5) The mechanisms responsible for this transition were seed predation and soil disturbance by kangaroo rats Preferentially eating the seeds of competitively dominant grasses, kangaroo rats indirectly released subordinate plant species from competition Furthermore, kangaroo rat burrowing favored disturbance-tolerant plant species Hence, the presence or absence of kangaroo rats determined whether the community was desert shrubland or arid grassland Note that this example both extends the concept to herbivores and attributes the keystone effect to a multispecies group instead of a single species Recently Brown (1998) suggested that the kangaroo rat effect was attributable primarily to two species, Dipodomys merriami and Dipodomys spectabilis D spectabilis went locally extinct in 1994, so if the system persists unchanged well beyond this extinction, D merriami may be a keystone species In this study, however, the original interpretation of these herbivores as a keystone ‘‘guild’’ may contribute Figure Densities of perennial and annual grasses in the presence and absence of kangaroo rats (Dipodomys) Data are mean percent cover (71 SE) of Eragrostris lehmanniana (ERIN), other tall perennial grasses (PERG), Aristida adscensionis (ARAD), other tall annual grasses (ANNG), and the short annual grasses Bouteloua aristidoides (BOAR) and Bouteloua barbata (BOBA) Top: Data from transects inside enclosures from which kangaroo rats had been removed or allowed to remain Asterisks above bars indicate significant differences between ỵ kangaroo rat and À kangaroo rat enclosures Bottom: Data from transects outside the respective enclosures Asterisks between panels indicate significant differences between transects inside and outside plots where kangaroo rats were present or absent Analyses employed ANOVA with plots as units of replication; Ã, po.0.05; ÃÃ, po0.01; ÃÃÃ, po0.001 Reproduced from Brown JH and Heske EJ (1990) Control of a desert–grassland transition by a keystone rodent guild Science 250: 1705–1707, with permission from American Association for the Advancement of Science to confusion in terminology This study is an excellent example of the strong effects of consumers on community structure, but as explained later, it was not designed to discern between keystone and ‘‘diffuse’’ (multispecies) predation Bison have also been identified as keystone herbivores (Knapp et al., 1999) Studies at Konza Prairie have documented strong effects of bison on species composition, diversity, and several physical and chemical aspects of ecosystem function of grassland communities These patterns occur across scales ranging from the plant level to patches to