Keystone predation can be a determinant of community structure, including species diversity, but factors underlying “keystoneness” have been minimally explored. Using the system in which the original keystone, the sea star Pisaster ochraceus, was discovered, we focused on two potential (but overlapping) determinants of keystoneness: intrinsic traits or state variables of the species (e.g., size, density), and extrinsic environmental parameters (e.g., prey productivity) that may provide conditions favorable for keystone predator evolution. Using a comparative-experimental approach, with repeated field experiments at multiple sites across a variable coastal environment, we tested predation rates, or how quickly predators consumed prey, and predation effects, or community response to predator presence or absence. We tested five hypotheses: (H₁) predation rates and effects will vary in space but not time; (H₂) per population predation rates will vary primarily with individual traits and population variables; (HJHH₃) per capita predation rates will vary only with individual traits; (H₄) predation effects will vary with traits, variables, and external drivers; and (H₅) as predicted by the keystone predation hypothesis, diversity will vary unimodally with predation pressure. As hypothesized, predation rates differed among sites but not over time (H₁), and in caging exclusion experiments, predation effect varied with both intrinsic and extrinsic factors (H₄). Unexpectedly, predation rates varied with both intrinsic and extrinsic (H₂, per population), or only with extrinsic (H₃, per capita) factors. Further, in large-plot exclusion experiments, predation effect was most closely associated with individual traits (contra H₄). Finally, taxon diversity varied unimodally with proxies of predation pressure (sessile prey abundance) and was sensitive to extrinsic factors (mussel growth, temperature, and upwelling) (H₅). Hence, keystoneness depended on predator individual traits, predator population variables, and environmental parameters. However, temporal differences in caging experiments suggested that environmental characteristics underlying prey dynamics may be preeminent. Compared to prior experiments, predation was weaker with low prey input compared to periods with high prey input. Collectively, our results suggest that keystone predator evolution depends on the coalescence of species-specific characteristics, and environmental parameters favoring high prey productivity. Our approach may be a model for future studies exploring the generality of keystoneness.

Data were collected using community surveys, field experiments testing predation effects and rates of predation, surveys of sea star density, sea star size structure, mussel and barnacle recruitment, mussel growth, bottle samples of nutrients and chlorophyll-a, and air and water temperature. Analysis used JMP, PRIMER-E, and PERMANOVA.