The Intermediate Disturbance Hypothesis

Disturbance is any relatively discrete event — a storm, fire, treefall, frost, predator irruption, flood, or human land-use event — that removes biomass, opens resource-rich patches, or kills organisms outright, thereby resetting local succession. Disturbance regimes are characterized by three interacting axes: frequency (how often disturbances occur, e.g., mean time between events), intensity (how severe each event is, e.g., proportion of biomass or cover removed), and spatial scale (the size of the patch affected relative to the dispersal range of the resident species). The Intermediate Disturbance Hypothesis, most fully articulated by Joseph Connell (1978) building on earlier ideas from Grime, Dayton, and others, states that local species diversity — usually measured as species richness, sometimes richness plus evenness — is maximized when disturbance frequency and/or intensity are neither very low nor very high. At low disturbance, communities approach a competitive equilibrium in which a small number of species with high competitive ability for the limiting resource dominate, driving losers to local extinction exactly as the The Competitive Exclusion Principle predicts. At high disturbance, only species with rapid colonization, fast growth, and high reproductive output — classic "weedy" or fugitive life histories — can persist between successive resets; slower-growing, more competitive species never get the uninterrupted time needed to establish. At intermediate disturbance, both early-successional colonizers and later-arriving competitors coexist, because the disturbance regime continually creates new open patches (favoring colonizers) while also leaving enough undisturbed time and space for competitive dominants to establish in older patches, before they, too, are eventually reset. Crucially, the hypothesis is fundamentally about non-equilibrium coexistence: diversity is sustained not because species reach a stable competitive balance, but because the community is a dynamic mosaic of patches at different stages of Ecological Succession, and different species dominate different patch-ages. No single patch supports the full diversity; the landscape does, because disturbance continually asynchronizes succession across patches. This is why IDH is properly a hypothesis about how disturbance regimes interact with successional dynamics and competitive hierarchies — not simply "disturbance is good for diversity," which overstates and oversimplifies the claim.

Step 1 — Baseline: undisturbed competitive sorting. In the complete absence of disturbance, species compete for shared limiting resources (light, space, nutrients, prey). Given enough time and a spatially uniform environment, the species with the lowest resource requirement at equilibrium (lowest R*, in resource-competition terms) or highest competitive ability displaces the others from that patch. Diversity within the patch collapses toward monodominance — the outcome the competitive exclusion principle predicts for a single limiting resource in a constant environment.

Step 2 — Disturbance interrupts the sorting process. A disturbance event removes some or all of the resident biomass in a patch, freeing resources (light gaps in a canopy, open substrate on a reef, bare soil in a grassland) and resetting that patch's successional clock to an earlier stage. Whether competitive exclusion is allowed to run to completion in any given patch depends on how long the interval between disturbances is relative to the time the slowest, most competitive species needs to establish and dominate.

Step 3 — Colonization versus competitive replacement — a race against time. Two functional guilds respond differently:

Colonizers arrive first in any newly opened patch (they are typically dispersal-limited but establishment-fast); competitors arrive more slowly but, once established, outcompete the colonizers for the limiting resource. This produces a predictable successional sequence within each patch — precisely the process described in Ecological Succession — that disturbance interrupts at different points.

Step 4 — Disturbance frequency determines which guild wins locally, and diversity depends on the mix.

• Low frequency / low intensity: Most patches age long enough for competitors to arrive and exclude colonizers. The landscape becomes dominated by a handful of superior competitors. Diversity is low.
• High frequency / high intensity: Patches are reset before competitors can arrive or establish. Only fast colonizers persist, and they persist everywhere. Diversity is again low — but for a different reason (a filter on life-history traits rather than competitive exclusion).
• Intermediate frequency/intensity: The landscape contains a mixture of patch ages — some recently disturbed (favoring colonizers), some mid-successional (colonizers and early competitors coexisting), some long undisturbed (dominated by top competitors, but not yet reset). Because different species dominate different patch ages, and disturbance continually creates new patches faster than any one competitor can spread across the whole landscape, many species coexist regionally even though few coexist within any single patch. This spatial and temporal patchiness is the heart of the mechanism.

Step 5 — The role of spatial scale and dispersal. IDH's predictions depend on disturbed patches being small and frequent enough, relative to organisms' dispersal ranges, that colonizers can reliably reach new gaps, and undisturbed patches lasting long enough, relative to competitors' growth rates, for exclusion to nearly — but not completely — occur across the whole landscape. If patches are disturbed synchronously across the entire region (or if dispersal is too limited to recolonize), the mosaic collapses and diversity falls regardless of mean disturbance frequency.

Connell (1978) — coral reefs and rainforests. Joseph Connell's landmark paper, "Diversity in Tropical Rain Forests and Coral Reefs" (Science, 1978), synthesized field observations from both systems. On coral reefs subject to periodic cyclone damage and on rainforest plots subject to variable treefall gap frequency, Connell noted that the most diverse patches of forest or reef were consistently those with a history of moderate, intermediate disturbance — not the areas that had gone longest without disturbance (which tended toward dominance by a few long-lived, shade-tolerant, or slow-growing competitive species) nor the areas disturbed most frequently or severely (which were dominated by a few fast colonizers). Connell explicitly framed this as evidence against a simple equilibrium view of tropical diversity, and it became the founding statement of IDH.

Paine's keystone predation experiments — a related non-equilibrium demonstration. Robert Paine's rocky intertidal work (starting 1966, later including sea urchin and Pisaster removal experiments in the 1970s) is often cited alongside IDH because it demonstrates the same underlying logic — that removal of biomass (via predation, functioning like a biotic disturbance) prevents monopolization by a dominant competitor. When Paine removed the predatory starfish Pisaster ochraceus from experimental plots on the Washington coast, mussel (Mytilus californianus) populations expanded and monopolized the substrate, driving local species richness from roughly fifteen species down to eight within about a year, as barnacles, chitons, and algae lost space to competitively superior mussels. This showed directly that removing a disturbance-like force (predation-driven mortality) allowed competitive exclusion to proceed and diversity to collapse — the mirror image of the IDH prediction.

Grassland and old-field mowing/grazing experiments. Controlled manipulations of grazing or mowing frequency in temperate grasslands (e.g., studies following on from Grime's work on plant strategies and from long-term grassland disturbance trials) have repeatedly found hump-shaped richness responses to disturbance frequency: ungrazed or unmown plots become dominated by a few tall, competitive grasses that shade out smaller species; heavily grazed or frequently mown plots are dominated by a few grazing-tolerant, prostrate species; plots at intermediate grazing/mowing intensity retain the highest number of coexisting plant species, because periodic removal of dominant biomass opens light and space for subordinate species without eliminating slower-establishing species entirely.

What the combined evidence does and does not prove. These studies establish that (1) disturbance interacts with competitive hierarchies to shape diversity, (2) removing disturbance allows competitive exclusion to proceed toward the compositions the competitive exclusion principle predicts, and (3) intermediate disturbance regimes are frequently — though not universally — associated with peak richness. They do not prove that a hump-shaped curve is a universal law; subsequent meta-analyses (notably a 2004 review by Mackey and Currie surveying hundreds of published disturbance–diversity relationships) found that only a minority of empirical datasets show a clean hump-shaped pattern, with many showing monotonic increases, decreases, or no significant relationship. This has led the field to treat IDH as one important mechanism among several, not a universal rule — a nuance addressed further under misconceptions below.

Example 1 — A quantitative succession/disturbance model. Imagine a rocky shore divided into 100 equal patches. After a disturbance clears a patch, colonization proceeds through a fixed successional sequence:

• Years 0–2: bare patch colonized by fast-growing green algae (colonizer guild).
• Years 2–6: green algae are gradually replaced by a mix of algae and slower-growing barnacles/mussels (mixed guild — highest local richness).
• Years 6+: mussels competitively exclude nearly everything else, forming a monoculture (competitive dominant, low richness).

Suppose disturbances strike patches randomly and independently, with a mean return interval of T years per patch (so in any year, the probability a given patch is disturbed is roughly 1/T).

• If T = 1 year (very frequent disturbance): nearly every patch is reset before year 2, so almost all patches sit in the "bare/colonizer" stage. Regional richness ≈ richness of the colonizer guild only (say, 3 algal species). Low diversity.
• If T = 30 years (very infrequent disturbance): the vast majority of patches have long since passed year 6 and become mussel monoculture. Regional richness ≈ richness of the dominant guild only (say, 2 mussel/barnacle species). Low diversity.
• If T = 4 years (intermediate): using a simple stationary age-distribution approximation, the fraction of patches in each age-class is proportional to the time spent in that class before disturbance resets it (roughly, patches are uniformly spread across ages 0 to T). With T = 4, patches are spread across ages 0–4, meaning many patches fall in the 2–6 year "mixed" window where richness is highest, and few have survived long enough to become monocultures. Regional richness sums species from all three guilds actively present in the mosaic — call it 3 (colonizers, still present in the youngest patches) + 4 (mixed-stage species, algae and juvenile barnacles/mussels co-occurring) + 0 (essentially no patches old enough for full mussel dominance) = 7 species, higher than either extreme.

This illustrates the core quantitative logic: regional richness is the union of species present across the standing distribution of patch ages, and that union is maximized when the disturbance return interval is comparable to the time required to pass through — but not complete — the full successional sequence.

Example 2 — Reasoning through a real management scenario (grazing intensity in a savanna reserve). A wildlife reserve manager wants to set a controlled burn frequency in a mixed grassland-savanna to maximize plant species richness for grazing wildlife. Historical records show:

• Under fire suppression (no burns for 20+ years), the savanna converts to closed woodland dominated by a few fire-intolerant tree species, understory grass richness drops to roughly 8 species per hectare (light-competition exclusion).
• Under annual burning, only a handful of highly fire-tolerant, fast-resprouting grasses persist — richness roughly 10 species per hectare (a strong filter on fire tolerance).
• Under burning roughly every 3–5 years, richness rises to roughly 35 species per hectare: enough fire-free time for many forbs and slower grasses to establish and flower, but frequent enough burning to prevent woody encroachment and prevent any single grass from forming a closed sod that excludes subordinate species.

Applying the mechanism: fire functions as the disturbance; woody encroachment (light competition) is the competitive-exclusion endpoint being interrupted; fire-tolerant resprouters are the colonizer/tolerator guild. The manager should set burn frequency near the 3–5 year interval — matching the time scale over which competitive dominance by trees would otherwise become established but before it does — rather than at either extreme, and should also stagger burns spatially (patch-burning rather than burning the whole reserve synchronously) so that a mosaic of post-fire ages is maintained across the landscape at any one time, replicating the spatial-mosaic mechanism that makes IDH work.

Joseph H. Connell (1923–2020) was an Australian-American ecologist whose long-running field experiments in intertidal and coral reef ecology, and later in tropical rainforest plots in Queensland, made him one of the most influential empirical community ecologists of the twentieth century. Connell was known for insisting on manipulative field experiments over armchair theorizing — his barnacle competition studies on the Scottish coast in the 1950s and 60s (showing Balanus outcompeting Chthamalus for space) had already demonstrated that competitive exclusion in nature is often incomplete because of environmental disturbance and predation, laying the groundwork for his later synthesis. His 1978 Science paper crystallized decades of reef and rainforest observation into the concise, testable hump-shaped prediction that became known as the Intermediate Disturbance Hypothesis, and he continued to publish influential critiques and refinements of successional and diversity theory into the 1990s.

Conservation and habitat management. Fire ecology, prescribed burning regimes, and rotational grazing schedules in grasslands, chaparral, and savanna reserves are directly designed using IDH logic: managers deliberately maintain intermediate, patchy disturbance regimes (rather than total fire suppression or uncontrolled burning) to sustain plant and, indirectly, animal diversity, echoing the reasoning in Example 2 above.

Forestry. Selective logging and small-gap silviculture practices are informed by the recognition that moderate, patchy canopy disturbance (mimicking natural treefall gaps) sustains higher tree and understory diversity than either clear-cutting (too intense/synchronous) or complete harvest exclusion (allows canopy dominants to monopolize light).

Coral reef management. Reef managers use IDH-derived reasoning when assessing the effects of storm frequency, bleaching events, and human physical damage (anchoring, trampling) on reef diversity, and when designing marine protected area disturbance regimes (e.g., permitting limited, rotational access rather than either total closure or unrestricted use).

Agriculture and rangeland science. Rotational and mixed-intensity grazing systems are explicitly designed around intermediate disturbance logic to maintain forage plant diversity and soil health, avoiding both overgrazing (which favors only grazing-tolerant weeds) and undergrazing (which allows a few dominant grasses or woody species to take over).

Restoration ecology. Restoration practitioners reintroduce disturbance (mowing, controlled burns, mechanical soil disturbance) into degraded or overly stable ecosystems specifically to interrupt competitive dominance by invasive or weedy monocultures and re-open opportunities for native species establishment — a direct, applied test of the mechanism described above.

Human and microbial ecosystems. The logic has been extended (with caveats) to gut microbiome disturbance from antibiotics or diet change, forest pest outbreaks, and even to models of disease-driven disturbance in wildlife populations, illustrating how broadly the coexistence-via-disturbance mechanism generalizes beyond its original plant and reef contexts — while underscoring, again, that its applicability must be tested case by case rather than assumed.

Ask the tutor: How would you predict the shape of the diversity–disturbance curve differ between a fast-growing phytoplankton community and a slow-growing old-growth forest, given their very different generation times?