Keystone Species

A keystone species is a species whose impact on the structure, diversity, or function of its community is disproportionately large relative to its abundance, biomass, or productivity. The concept was named by Robert Paine in 1969, building directly on his own removal experiments on the sea stars Pisaster ochraceus. The crucial word is disproportionate: a keystone species is not simply an important species or a dominant one. Community dominants — the most abundant plant in a grassland, the most numerous fish in a lake — matter a great deal, but their importance scales roughly with how much of the community they constitute. A keystone species breaks that proportionality: a rare, low-biomass actor exerts effects far beyond what a simple census would predict, usually by controlling the population of a competitively dominant species lower or elsewhere in the web. The concept sits at the intersection of two ideas students already know. From The Lotka-Volterra Predator-Prey Model comes the basic mechanism — predation regulates prey density — but the classic model treats every predator-prey pair as dynamically generic, with no sense that removing this predator could restructure an entire multispecies assemblage. From the study of Mutualism and Commensalism comes the recognition that species interactions can create dependencies that ripple outward — a fig tree's fruit sustaining a whole guild of frugivores is a mutualism-flavored keystone effect, distinct from Paine's predation-flavored one. Keystone species theory generalizes both: what matters is not the type of interaction (predation, mutualism, competition, engineering) but its topological position in the web — is it a chokepoint through which a disproportionate share of community-level consequences flow? Ecologists now distinguish several flavors of keystone role, though the core logic — a small cause with a large, structure-altering effect — is shared across all of them:

A related but distinct term is foundation species — usually an abundant, habitat-forming organism (a coral, a kelp, a tree) whose importance does scale with its biomass, because it physically constitutes the habitat. Confusing the two is one of the most common errors students make, addressed below.

The clearest mechanistic story, and the one that gave the concept its name, is Paine's intertidal system on the outer coast of Washington State.

Step 1 — The competitive hierarchy exists but is normally suppressed. On wave-exposed rocky benches, the California mussel Mytilus californianus is the superior competitor for the primary limiting resource: physical space on the rock. Mussels grow in dense, multilayered beds, attach with strong byssal threads, and can overgrow and smother barnacles, algae, chitons, and anemones that settle nearby. Left unchecked, mussels would competitively exclude nearly everything else from the low-to-mid intertidal zone.

Step 2 — The predator selectively removes the dominant competitor. The sea star Pisaster ochraceus forages up from subtidal refuges at high tide and preys heavily and preferentially on mussels, along with barnacles and snails, but mussels are its single largest prey category by volume in this system. Crucially, Pisaster does not eat all species equally — it is size- and species-selective, cropping mussels at the edges of beds before the beds can spread and take over available substrate.

Step 3 — Predation opens space, and space is what everything else is competing for. Because free rock surface is the limiting resource for the whole sessile community, every bit of space Pisaster clears becomes available to barnacles, algae, anemones, chitons, and limpets that cannot otherwise establish a foothold against mussels. This is the mechanistic crux: the predator is not adding energy or nutrients to the system — it is redistributing the limiting resource by keeping the strongest space-competitor in check.

Step 4 — Removing the predator collapses the redistribution. Take Pisaster away, and mussels — freed from their one significant natural enemy in this zone — spread downward and laterally until they occupy essentially the entire suitable substrate, physically excluding the species that depended on the gaps Pisaster used to maintain. Species richness in the plot falls because the entire mid-intertidal switches from a mosaic of habitat patches to a mussel monoculture.

Step 5 — The effect is asymmetric and non-substitutable. No other predator in the system fills the gap; whelks and small crabs take some mussel spat but cannot control adult mussel cover the way Pisaster does. This asymmetry — one species doing a job that nothing else can do at comparable scale — is exactly what makes the role "keystone" rather than merely "one predator among several." A second, independently discovered case reveals the same logic acting through a herbivore-controlling predator rather than directly on space competition, and shows the mechanism generalizes across very different systems.

Sea otters and the kelp forest. Along the North Pacific coast, sea otters (Enhydra lutra) prey heavily on sea urchins. Urchins graze kelp holdfasts and can, if unchecked, deforest a kelp bed entirely, converting a three-dimensional kelp forest into a barren, low-diversity "urchin barren." Where otters are present, urchin populations are kept sparse and cryptic (hiding in crevices); kelp forests flourish, providing habitat for fish, invertebrates, and other mammals. Where otters were extirpated by the fur trade, urchin populations exploded and kelp forests collapsed into barrens over large stretches of coastline — a mechanistically distinct route (herbivore control, not direct space competition) to the same qualitative outcome: loss of a low-abundance predator restructures the entire community.

The kelp-forest case also illustrates a trophic cascade — a chain of indirect effects passing down (or up) a food chain — which is the dynamical consequence that follows once a keystone predator is added or removed; the general theory of these cascades is developed further under Trophic Cascades.

The setup. In 1963, Robert T. Paine, then at the University of Washington, selected study sites at Mukkaw Bay on the outer Washington coast — wave-exposed rocky benches in the mid-intertidal zone with a natural community that included the mussel Mytilus californianus, several barnacle species, a chiton, two limpet species, four algal species, an anemone, and the predatory sea star Pisaster ochraceus, among others — roughly 15 macroscopic species occupying a single rock face.

The manipulation. Paine physically removed every Pisaster he could find from an experimental plot, repeatedly, by prying them off the rocks and throwing them as far out to sea as he could — a manual, continuous removal maintained over multiple years — while an adjacent, otherwise identical control plot was left undisturbed. This is a textbook manipulative field experiment: a single variable (presence/absence of one predator species) was altered while everything else about the physical site was held constant, with a true control for comparison.

The observation. Within about a year, the removal plot showed a rapid takeover by Mytilus californianus and the barnacle Balanus glandula, followed by near-total dominance of the substrate by mussels as they spread and overgrew the barnacles as well. Species that had occupied the space now smothered by mussels — including several algae, an anemone, and grazing gastropods — disappeared from the plot entirely. Species richness in the removal plot fell from roughly 15 species to about 8, and continued to decline toward a near-monoculture of mussels as the mussel bed matured and thickened over subsequent seasons, while the control plot's diversity remained stable.

What it proved. The result demonstrated that a predator's community importance need not scale with its abundance or biomass: Pisaster was numerically minor in the system, yet its removal caused a measurable, reproducible, directional collapse in local species richness that did not occur in the untouched control. Paine formalized the finding in his 1966 paper "Food Web Complexity and Species Diversity" and coined the term in his 1969 paper "A Note on Trophic Complexity and Community Stability," where he explicitly proposed that some species function as a keystone in an arch — remove it, and the rest of the structure (the community) is far more likely to fail.

A parallel, independent confirmation: the sea otter story. The otter–urchin–kelp relationship was documented through a different kind of evidence — a natural experiment rather than a manual manipulation. Comparisons across Aleutian and Alaskan islands with and without otters (studied especially by James Estes and John Palmisano in the 1970s) showed the same qualitative pattern Paine had produced by hand: islands with otters had dense kelp forests and low urchin density, islands without otters had urchin barrens and almost no kelp. When otters recolonized or were reintroduced to some islands, urchin barrens reverted to kelp forest within a few years — the natural equivalent of Paine's removal-and-recovery design, and it converged on the same conclusion from a completely different taxonomic system. Later still, when otter populations in parts of the Aleutians collapsed again in the 1990s (attributed by Estes and colleagues to increased predation by killer whales, itself linked to depletion of the whales' usual prey by industrial whaling), urchin barrens re-expanded — a third independent replicate of the same causal chain, observed rather than manipulated. Together, the Pisaster removals and the otter comparisons illustrate the two main evidentiary routes ecologists use for community-level causation: controlled manipulation with a control plot, and natural experiments exploiting real-world variation in the presence of the putative keystone — with convergent results strengthening confidence that the causal story, not some confound, explains the pattern.

Example 1 — Quantifying "disproportionate": a simple keystone index. Suppose in an intertidal survey you record the relative biomass of each species in a plot and, from a set of small-scale removal or transplant trials, estimate each species' "community importance" as the fractional change in total local species richness caused by removing it. Paine (and later Mark Power and colleagues, 1996) formalized this intuition with a simple community importance index:

where d_N is a diversity/composition measure with species i present, d_D is the same measure after its removal, and p_i is species i's proportional abundance (biomass fraction). Suppose:

For Pisaster:

For Mytilus (imagine instead removing the mussel, which would also change richness by opening space, but its biomass fraction is far higher):

Even though removing the mussel causes a larger absolute drop in richness (15→3 versus 15→8), dividing by its abundance shows its effect is proportionate to how common it already is — an index near 1 is typical of an ordinary dominant. The sea star's index, ~23, is an order of magnitude higher: a tiny biomass fraction is producing an outsized structural effect. This numeric contrast is exactly what "keystone" is meant to flag — not the biggest effect in absolute terms, but the biggest effect per unit abundance.

Example 2 — Predicting the outcome of a keystone loss using a simple competitive-exclusion logic. A lake has three benthic algal grazing snail species (A, B, C) competing for the same limiting resource (attached algae on rock surfaces), plus one specialist predatory fish that preferentially eats snail species A, the fastest-growing and most competitively dominant of the three when predation-free. Without the fish, species A's intrinsic growth rate advantage lets it competitively exclude B and C within a few generations (a Lotka–Volterra-style competitive exclusion outcome, following logic parallel to the two-species case in The Lotka-Volterra Predator-Prey Model but here applied to competitors sharing a predator). Suppose the following approximate per-capita growth/mortality parameters apply, with the fish removing 40% of species A's standing population per generation but only 5% of B and 3% of C (because it is a specialist that only rarely takes the other species):

Net effective growth with fish present:

• A: 1.30 × (1 − 0.40) = 0.78/gen
• B: 1.05 × (1 − 0.05) = 1.00/gen
• C: 0.95 × (1 − 0.03) = 0.92/gen

With the fish present, species A's effective growth rate (0.78) drops below both B (1.00) and C (0.92) — predation converts the strongest competitor into the weakest, allowing all three species to coexist in a rough mosaic, exactly analogous to Pisaster converting mussels from unstoppable to manageable. Remove the fish, and A's raw growth advantage (1.30 versus 1.05 and 0.95) reasserts itself; over successive generations A should competitively exclude B and C, and local richness in that patch should fall from 3 species to 1 — a smaller-scale numerical replay of what happened to Paine's plots when Pisaster was removed. This worked example shows why keystone predation stabilizes diversity mechanistically: it is not the predator's presence per se, but its selective suppression of whichever competitor would otherwise dominate, that maintains the mosaic.

Robert Treat Paine III (1933–2016) was an American ecologist who spent almost his entire career at the University of Washington, working from the Friday Harbor Laboratories on the rocky shores of the Pacific Northwest. Trained initially in paleontology and invertebrate zoology, Paine turned to intertidal ecology in the early 1960s at a time when community ecology was dominated by descriptive natural history and by theoretical modeling in the Lotka–Volterra tradition, with relatively few people testing community-level hypotheses through direct, sustained field manipulation. His decision to physically and repeatedly remove a predator from real rocky-shore plots, rather than simply modeling or observing, was itself a methodological turning point: it treated the intertidal zone as a system amenable to controlled experimentation, much as a physiologist would treat a nerve preparation. Paine went on to mentor a lineage of experimental community ecologists (including Paul Dayton, Peter Kareiva conceptually adjacent, and others), and continued fieldwork at Tatoosh Island into his later career, tracking how Pisaster dynamics interacted with sea temperature and other stressors over decades — turning a single removal experiment into one of ecology's longest-running natural experiments.

Conservation prioritization. Wildlife agencies increasingly use keystone-species reasoning to decide where limited conservation funding produces the largest community-level payoff — protecting one low-abundance apex predator can preserve an entire assemblage more cost-effectively than protecting many species piecemeal. The reintroduction of gray wolves to Yellowstone National Park in 1995 is widely (though not without ongoing scientific debate about the magnitude of effects) framed in these terms, with proposed downstream effects on elk browsing behavior, willow and aspen regeneration, and even river channel geomorphology.

Marine reserve and fisheries design. Because removing a single predator (via overfishing or bycatch) can trigger disproportionate community collapse — as with sea otters and, in other systems, sharks controlling ray or grouper populations that in turn affect shellfish beds — marine protected area boundaries and catch limits are sometimes explicitly designed around protecting known or suspected keystone predators rather than simply maximizing total catch.

Restoration ecology. Sea otter reintroduction programs along the Pacific coast of North America have been used as a deliberate restoration tool: reintroducing the keystone predator, rather than trying to manage urchins or replant kelp directly, restores the community indirectly and more durably, because it restores the mechanism rather than just the current symptom.

Agricultural and applied ecology. The mutualist-flavored keystone concept informs pollinator conservation policy: because a small number of pollinator taxa can support disproportionate fractions of both wild plant reproduction and crop yield, pollinator protection (habitat strips, pesticide restrictions) is often justified using keystone-style cost–benefit logic even outside marine systems.

Disease ecology and invasive species management. Removing an invasive species that has become functionally dominant, or reintroducing a native predator suppressed by human activity, is now routinely evaluated using the same disproportion-to-abundance logic Paine developed — asking not just "how many are there" but "what structural role do they occupy," a question that leads naturally into the broader study of Food Webs and Community Structure and the propagation of Trophic Cascades through entire ecosystems.

Ask the tutor: How would the keystone concept apply if two predator species partially, but not fully, overlap in the prey they suppress — would losing one still cause community collapse?