27.01.04 · earth-science / plate-tectonics

Mantle plumes, hot spots, and large igneous provinces: Hawaii, Yellowstone, and the Deccan Traps

shipped3 tiersLean: none

Anchor (Master): Wilson 1963 Can. J. Phys. 41:993; Morgan 1971 Nature 230:42; Richards-Duncan-Ball 1989; Campbell-Griffiths 1990; Courtillot 2003; Foulger 2005

Intuition Beginner

Most volcanoes sit at plate boundaries. The Pacific Ring of Fire traces subduction-zone volcanoes around the Pacific rim; Iceland sits on the Mid-Atlantic Ridge where plates pull apart. But some volcanoes appear in the middle of plates, far from any boundary. Hawaii, in the middle of the Pacific Plate, is the canonical example. In 1963, the Canadian geophysicist Tuzo Wilson proposed that these "hot spots" are fixed points in the deep mantle.

Wilson's picture: a column of hot solid rock, called a mantle plume, rises from the core-mantle boundary, about 2,900 kilometers down. The plume melts partially as it nears the surface, supplying magma through the plate above. As the plate drifts over a plume that stays fixed in the mantle, a chain of volcanoes forms, older extinct ones on one side and the active volcano on top.

The Hawaiian-Emperor chain is a 6,000-kilometer line of volcanoes crossing the Pacific. It records more than 70 million years of Pacific Plate motion over the Hawaiian plume. The sharp bend in the chain, about 43 million years old, records a change in the direction of plate motion. Hot-spot chains are the only tool geologists have for reconstructing absolute plate motion through geologic time.

Visual Beginner

Feature	Length / volume	Age	What it records
Hawaiian-Emperor chain	~6,000 km of volcanoes	0 to 80 Ma	Pacific Plate motion over the Hawaiian plume
43 Ma bend	kink in the chain	~43 Ma	Change in plate-motion direction
Deccan Traps (LIP)	~1.5 million km³ of lava	~66 Ma	Plume-head arrival under India (Reunion)
Siberian Traps (LIP)	~4 million km³ of lava	~252 Ma	Plume-head arrival; P-T extinction
Yellowstone-Snake River Plain	~800 km track	~16 Ma to 0	North American Plate motion over Yellowstone plume

Worked example Beginner

About 66 million years ago, India was crossing the Indian Ocean, far from Asia, and the Reunion hot spot sat beneath it. Over roughly 1 million years, the plume erupted 1.5 million cubic kilometers of lava, covering about 1.5 million square kilometers of India in layers up to 2 kilometers thick. Geologists call this vast lava outpouring a flood basalt, and the regional unit a large igneous province, or LIP.

Step 1. Radiometric dating of Deccan lava flows gives ages centered near 66 million years ago. The eruptions occurred in pulses spanning about 1 million years.

Step 2. The Chicxulub asteroid impact, which marks the Cretaceous-Paleogene (K-Pg) boundary and the end of the non-avian dinosaurs, dates to 66 million years ago. The Deccan eruptions and the impact overlap in time.

Step 3. The end-Cretaceous mass extinction extinguished about 75 percent of plant and animal species, including all non-avian dinosaurs. Geologists debate the relative contributions of the Deccan volcanism and the impact: the volcanism stressed ecosystems for hundreds of thousands of years before the impact delivered the final blow.

What this tells us: when a plume head reaches the base of a plate, it can release staggering volumes of lava in geologically short time, and that volcanism correlates with mass extinction. The Siberian Traps, at 252 million years ago, align with the Permian-Triassic extinction, the "Great Dying" that killed roughly 96 percent of marine species.

Check your understanding Beginner

Exercise 1 (easy, multiple choice).

Where do most hot-spot volcanoes form?

A. At subduction-zone boundaries, where one plate dives beneath another

B. At mid-ocean ridges, where plates pull apart

C. In the interior of a plate, far from any boundary

D. Only at transform faults, where plates slide horizontally

Hint

Recall where Hawaii, the canonical hot-spot example, sits relative to the Pacific Plate.

Answer

Option C. Hot spots are defined by their occurrence in plate interiors, far from any boundary. Hawaii sits roughly in the middle of the Pacific Plate, thousands of kilometers from the East Pacific Rise and from the subduction zones of the Pacific Ring of Fire. Subduction-zone and ridge volcanoes (options A and B) have a different origin: water-release melting and decompression melting at boundaries, respectively.

Formal definition Intermediate+

A mantle plume is a buoyant column of hot solid rock rising through the mantle, driven by thermal convection. A hot spot is the surface volcanic expression of a plume, including the active volcano and its trail of extinct predecessors. A large igneous province (LIP) is a flood-basalt province covering more than $1 0^{5} k m^{2}$ , with a volume exceeding $1 0^{5} k m^{3}$ , emplaced in less than a few million years.

Plume architecture (Morgan 1971; Campbell-Griffiths 1990)

The canonical plume-head/tail model ^{[Morgan 1971]} ^{[Campbell-Griffiths 1990]} is:

Plume head. A large bulbous thermal anomaly at the top of the plume, with diameter $800$ to $1, 200 km$ at the depth where it flattens against the lithosphere. The head forms as the plume ascends through the mantle, entraining ambient mantle and growing wider.
Plume tail. A narrower conduit, $100$ to $200 km$ in diameter, connecting the head to its source region, typically the core-mantle boundary at depth $2, 891 km$ , or in some models the transition zone at $660 km$ .
Head arrival. The plume head, on reaching the base of the lithosphere, undergoes decompression melting, producing $1 0^{6}$ to $1 0^{7} k m^{3}$ of basaltic magma in $1 0^{5}$ to $1 0^{6}$ years. The result is a continental flood basalt or an oceanic plateau: a LIP.
Tail phase. After head emplacement, the steady tail sustains a hot-spot volcano chain as the plate drifts overhead, with magma flux of $1 0^{2}$ to $1 0^{3} m^{3} /s$ .

Evidence categories for a plume (Courtillot 2003)

Courtillot and colleagues in 2003 ^{[Courtillot 2003]} proposed a five-criteria test for distinguishing "true" plumes from other volcanic phenomena:

Age progression. A linear chain of volcanoes with monotonically increasing ages along the chain, at a rate consistent with local plate motion.
Hot-spot reference-frame stationarity. When reconstructed against other hot spots, the candidate plume has remained approximately fixed in latitude and longitude.
High $^{3} He /^{4} He$ ratio. Helium-3 is primordial, retained in the deep mantle since Earth's accretion; high $^{3} He /^{4} He$ in plume basalts ( $R / R_{A} > 8$ , where $R_{A}$ is the atmospheric ratio) signals a deep-mantle source distinct from mid-ocean-ridge basalts ( $R / R_{A} \approx 8$ ).
High heat flow. A thermal anomaly at the surface above the plume.
Tomographic evidence. A seismically slow column extending through the lower mantle to the CMB, imaged by seismic tomography.

A plume candidate that satisfies all five is classified as a "primary" plume; one satisfying three or four is "secondary"; fewer is "doubtful." Of the roughly 50 catalogued hot spots, Courtillot and colleagues identify 7 to 9 primary plumes: Hawaii, Iceland, Afar, Reunion, Louisville, Easter, Galapagos, Kerguelen, Tristan.

Large igneous provinces and the LIP-extinction correlation

LIP	Age (Ma)	Volume (km³)	Extinction / boundary
Siberian Traps	~252	~4 million	Permian-Triassic (96% marine species)
CAMP (Central Atlantic Magmatic Province)	~201	~2 million	Triassic-Jurassic
Ontong Java Plateau	~120	~100 million	Oceanic anoxic event; minor extinction
Deccan Traps	~66	~1.5 million	K-Pg (75% species; Chicxulub impact coeval)
Columbia River Basalts	~16	~0.17 million	No major extinction; youngest LIP

The LIP-extinction correlation is strongest at the Permian-Triassic and Triassic-Jurassic boundaries, where no coincident asteroid impact is implicated. The K-Pg case is contested: the Chicxulub impact is the leading cause, with the Deccan volcanism contributing ecosystem stress before and after.

Yellowstone and the Snake River Plain

Yellowstone, the active caldera in northwestern Wyoming, is the present-day surface expression of a hot-spot track extending $800 km$ southwest across the Snake River Plain. The track records about $16 Myr$ of North American Plate motion over the Yellowstone plume, with successive caldera-forming eruptions progressively younger to the northeast. The most recent Yellowstone supereruption, $631, 000$ years ago, ejected $1, 000 k m^{3}$ of ash, blanketing much of the western United States.

The plume-vs-no-plume debate

The Foulger-Anderson school ^{[Foulger 2005]} argues that many proposed plumes fail criteria 1, 3, and 5 simultaneously, and that a shallow-mantle origin is more parsimonious. In their model, called the Plate model, extensional stress within the lithosphere lowers the melting pressure locally, generating magma without a deep plume. The observed volcanic chains are then interpreted as lithospheric fractures propagating across the plate, not as the surface expression of a fixed deep source.

The strongest Foulger-Anderson objections:

Several hot-spot chains lack a clean age progression (e.g., the Cameroon line).
The $^{3} He /^{4} He$ signal is not unique to a deep source; it can be concentrated in mantle domains that are simply ancient, not necessarily deep.
Tomographic images beneath Hawaii show a slow column to roughly $1, 000 km$ depth but lose resolution in the lower mantle, leaving the CMB connection unproven.

Counterexamples to common slips Intermediate+

"Hawaii sits at a plate boundary." It does not. Hawaii is in the interior of the Pacific Plate, more than $3, 000 km$ from the East Pacific Rise and more than $5, 000 km$ from the nearest subduction-zone trench. Its setting is the canonical case of a mid-plate hot spot.
"Hot spots are perfectly fixed." They drift relative to one another at $10$ to $50 mm/yr$ , accumulating hundreds of kilometers of misfit over $100 Myr$ . The hot-spot reference frame is an excellent first approximation, not an exact frame.
"All flood basalts come from plume-head arrival." Most do, but the Columbia River Basalts ( $\sim 16 Ma$ , the youngest LIP) lack a clear plume-head precursor and may instead reflect asthenospheric flow around the edge of the subducting Juan de Fuca slab.
"The plume hypothesis is settled." It is not. The Foulger-Anderson critique has serious arguments, particularly regarding the depth resolution of seismic tomography and the failure of some hot-spot tracks to satisfy Courtillot's age-progression criterion.
"LIPs cause all mass extinctions." Most LIPs correlate with extinction events, but the K-Pg extinction is most strongly tied to the Chicxulub impact. LIPs are the primary extinction driver at the Permian-Triassic (Siberian Traps) and likely at the Triassic-Jurassic (CAMP).
"The Hawaiian plume is anchored at 2,900 km depth." Probably, but the depth resolution of seismic tomography in the lower mantle is debated. The 2009 Wolfe et al. experiment ^{[Wolfe 2009]} imaged the Hawaiian plume as a slow seismic column extending into the lower mantle, but the CMB connection remains inference, not direct image.

Key model: the fixed-hot-spot reference frame Intermediate+

Model. Assume mantle plumes are approximately fixed relative to the lower mantle and to one another, with inter-plume drift of order $10$ to $50 mm/yr$ . Then the chain of volcanoes left by a plume on the overriding plate records the absolute motion of that plate through geologic time. The set of plumes defines a hot-spot reference frame. Given a hot-spot track with measured ages $t_{i}$ at positions $x_{i}$ , the plate velocity is

v_{plate} = \frac{d x}{d t}_{track},

and integrating the velocity of every plate against the frame yields a global reconstruction of absolute plate motion.

Key result: Hawaiian-Emperor age progression

The age of Hawaiian-Emperor volcanoes increases monotonically from the active island of Hawaii ( $0 Ma$ ) northwest along the chain to the Detroit Seamount near the Kuril Trench ( $\sim 80 Ma$ ). The linear fit of distance along the chain against radiometric age gives a Pacific Plate velocity of about $70 mm/yr$ for the past $43 Myr$ and about $50$ to $70 mm/yr$ before the bend, with the bend itself recording a roughly $6 0^{\circ}$ change in plate-motion direction at $43 Ma$ ^{[Wilson 1963]}.

Cross-hot-spot consistency

Reconstructions against multiple hot spots (Hawaii, Reunion, Louisville, Iceland) give concordant absolute plate-motion estimates to within $10$ to $50 mm/yr$ over the past $80 Myr$ . The concordance degrades for times older than about $120 Ma$ because plume-head phases disrupt stationarity: a young plume lacks the long tail needed to define a stable reference.

The reference-frame noise budget

The reference frame carries three noise sources:

Inter-plume drift ( $10$ to $50 mm/yr$ ). Plumes are advected by large-scale mantle flow and are not absolutely fixed.
Plume-head reorganization ( $> 50 mm/yr$ during head arrival). When a new plume head reaches the lithosphere, the surface position of the volcano can jump by hundreds of kilometers.
Plate-circuit closure errors ( $5$ to $10 mm/yr$ ). Plate motions are chained through relative-motion rotations about Euler poles across a global circuit, and small errors accumulate.

Putting these together, the hot-spot reference frame is accurate to roughly $10 mm/yr$ for the past $80 Myr$ and degrades for older times.

Defending the assumption

Three lines of evidence support the approximation:

Cross-hot-spot consistency (above). Multiple independent tracks give concordant plate velocities.
Plate-circuit closure. When the relative motion of the Pacific, African, and North American plates is reconstructed through ridge-transform geometry and matched to the hot-spot frame, the closure errors are at the $10 mm/yr$ level.
Tomographic plume images. The 2009 Wolfe et al. Hawaiian plume tomography ^{[Wolfe 2009]} shows a slow seismic column consistent with a deeply rooted plume, supporting the fixed-in-the-mantle assumption.

Derivation. Consider two hot spots $A$ and $B$ on plate $P$ . Let $v_{A} (t)$ and $v_{B} (t)$ be the absolute velocities of the hot spots in the lower-mantle frame. If $v_{A} \approx v_{B} \approx 0$ (the fixed-plume assumption), then the tracks left by $A$ and $B$ on the plate both record $- v_{P} (t)$ , the negative of the plate's absolute velocity. The distance between the two tracks at any time $t$ equals the fixed separation between the two plumes in the lower mantle, and is constant in time. Conversely, if the inter-plume distance $∥ x_{A} (t) - x_{B} (t)∥$ drifts, the rate of that drift measures the failure of the assumption. The observed drift is $10$ to $50 mm/yr$ , small compared to typical plate velocities ( $50$ to $100 mm/yr$ ). $□$

Bridge. The hot-spot reference frame builds toward 27.01.01 plate-tectonics survey, where the global map of plates and relative motions becomes absolute motion through the addition of plume tracks, and appears again in 27.08.01 geologic time scale, where the radiometric ages of hot-spot volcanoes calibrate the magnetic-anomaly timescale across the Pacific. The foundational reason plume tracks are the load-bearing tool for absolute plate motion is that sea-floor older than about $180 Ma$ has been subducted, destroying the marine magnetic record, while the hot-spot track on the overriding plate survives; this is exactly the gap that the Hawaiian-Emperor chain fills for the Pacific, the Walvis Ridge for the Atlantic, and the Ninetyeast Ridge for the Indian Ocean. The model generalises to every plate that has crossed a plume in the past $200 Myr$ , and the bridge is the recognition that without plumes, geophysicists would have no fixed reference for absolute plate motion older than the oldest preserved sea-floor.

Exercises Intermediate+

Exercise 4 (medium, multiple choice).

The bend in the Hawaiian-Emperor chain, dated at about $43 Ma$ , records:

A. A change in the position of the Hawaiian plume relative to the lower mantle

B. A change in the direction of motion of the Pacific Plate

C. The arrival of a new plume head at the surface

D. The subduction of the Emperor end of the chain

Hint

Under the fixed-plume hypothesis, the plume stays put. What can change?

Answer

Option B. Under the fixed-plume hypothesis, the bend records a roughly $6 0^{\circ}$ change in the direction of Pacific Plate motion at about $43 Ma$ . The change is thought to reflect a reorganization of Pacific Plate boundaries, possibly linked to the collision of India with Asia and the resulting reorientation of global plate circuits. The plume is assumed fixed; the bend is plate-motion geometry, not plume drift.

Exercise 6 (medium, short answer).

Explain in three or four sentences how the Foulger-Anderson "Plate" model accounts for an age-progression volcanic chain without invoking a deep mantle plume.

Hint

Consider a crack propagating across a plate: as the crack tip advances, it triggers decompression melting at progressively different points along its path.

Answer

In the Foulger-Anderson Plate model, a propagating lithospheric fracture or zone of extension moves steadily across a plate. As the crack tip advances, it locally lowers the pressure in the asthenosphere beneath, triggering decompression melting at progressively new positions along the path. The resulting volcanic chain has an age progression, but the source of the magma is shallow asthenosphere, not a deep plume. The geometry of the crack, governed by plate stress and prior tectonic history, controls the orientation of the chain.

Exercise 7 (medium, short answer).

Why is high $^{3} He /^{4} He$ ratio in plume basalts interpreted as evidence for a deep-mantle source?

Hint

Helium-3 is primordial; helium-4 accumulates over time from uranium and thorium decay.

Answer

Helium-3 was trapped in Earth during accretion and has no significant ongoing production. Helium-4 is produced continuously by the alpha-decay of uranium and thorium in the crust and mantle. Over geologic time, the $^{3} He /^{4} He$ ratio in any mantle reservoir that is well-mixed or chemically evolved decreases. Mid-ocean-ridge basalts (MORB), which sample the shallow upper mantle, have $R / R_{A} \approx 8$ , reflecting this evolution. Plume basalts with $R / R_{A} > 20$ (Hawaii, Iceland) require a reservoir that has been isolated from U-Th decay for billions of years, which in the standard picture is the deep lower mantle, preserved as a less-degassed reservoir distinct from the convecting upper mantle. The geochemistry therefore points to a deep-mantle source for the helium, supporting the plume hypothesis.

Exercise 8 (hard, short answer).

The Courtillot 5-criteria test classifies Hawaii as a primary plume but the Columbia River Basalts as a doubtful one. Identify which criteria Columbia River likely fails, and explain the implication for the origin of the Columbia River Basalts.

Hint

Consider whether the Columbia River Basalts show a clean age-progression track extending into the present, and whether a deep seismic anomaly has been imaged beneath them.

Answer

The Columbia River Basalts likely fail (i) age progression, since the eruptions occurred in a geologically brief interval around $16 Ma$ with no clear linear chain of progressively younger volcanoes extending to the present, (ii) hot-spot reference-frame stationarity, since there is no continuous Yellowstone-type track leading from the Columbia River to a present-day hot spot (the modern Yellowstone caldera sits at the end of the Snake River Plain trend, but the connection to the Columbia River flood basalt itself is discontinuous), and (iii) tomographic evidence, since no deep lower-mantle seismic anomaly has been imaged beneath the northwestern United States comparable to the Hawaiian plume. The implication is that the Columbia River Basalts may not have a plume-head origin. Alternative explanations include asthenospheric flow around the edge of the subducting Juan de Fuca slab, decompression melting behind the slab edge, or lithospheric extension; these mechanisms are consistent with the Foulger-Anderson Plate model.

Exercise 9 (hard, numeric).

Reconstruct the absolute plate-motion noise from inter-plume drift. Suppose two hot spots, A and B, are observed to drift apart at $30 mm/yr$ relative to each other. If a $100 Myr$ plate reconstruction uses both hot spots as fixed reference points, how far apart will the two reconstructed plate positions be at the start of the reconstruction, assuming the drift accumulated linearly in time? Give the answer in kilometers.

Hint

Multiply the drift rate by the time, converting millimeters to kilometers ( $1 km = 1 0^{6} mm$ ).

Answer

$3, 000 km$ . Drift rate $30 mm/yr$ over $100 Myr$ gives a total displacement of $30 \times 1 0^{8} mm = 3 \times 1 0^{9} mm = 3, 000 km$ . This is roughly one-third of the diameter of the Pacific Plate and reflects the limit of the hot-spot reference frame: reconstructions older than about $80$ to $100 Ma$ accumulate inter-plume drift of thousands of kilometers, degrading absolute-plate-motion estimates. For times older than the Cretaceous, geophysicists must rely on the paleomagnetic apparent polar wander path, which has its own noise.

Advanced results Master

Dana 1890: the Hawaiian deep-seated source

James Dwight Dana, geologist on the United States Exploring Expedition of 1838-1842, recognized that the Hawaiian volcanoes formed a linear progression and attributed them to a deep-seated source. His 1890 monograph Characteristics of Volcanoes ^{[Dana 1890]} argued, against the prevailing view that volcanoes were shallow features fed by a globally connected magma ocean, that Hawaiian volcanism required a sub-lithospheric source fixed beneath the moving Pacific plate. Dana's "hot spot" was not named as such, but his identification of the linear Hawaiian trend as a coherent geological feature is the first recognition of the hot-spot phenomenon.

Wilson 1963: the hot-spot hypothesis

J. Tuzo Wilson, working on the geometry of the Hawaiian Islands, proposed in 1963 ^{[Wilson 1963]} that the islands formed sequentially as the Pacific Plate drifted northwest over a fixed magma source. Wilson's Canadian Journal of Physics paper, "A possible origin of the Hawaiian Islands," used the increase in erosion and subsidence of the islands to the northwest to argue for age progression, then computed the implied plate velocity. Wilson was writing three years before the formal plate-tectonics revolution of 1967-1968, and his hot-spot hypothesis was a foundational contribution, providing the first absolute reference frame.

Morgan 1971: the founding plume paper

W. Jason Morgan in 1971 ^{[Morgan 1971]} formalized Wilson's intuition as the mantle plume hypothesis: thermal convection plumes rise from the lower mantle, fixed relative to one another, and produce hot-spot volcanism at the surface. Morgan identified about 20 global hot spots and proposed that their relative fixity defines a mantle-fixed reference frame against which plate motions can be measured. The Morgan paper, two pages in Nature, is the founding document of plume geophysics.

Crough 1978 and Duncan 1981: dating the tracks

S. Thomas Crough in 1978 showed that bends and kinks in hot-spot chains record changes in plate-motion direction, not plume drift. Robert Duncan and colleagues in 1981 applied high-precision radiometric dating to the Hawaiian-Emperor, Louisville, and Walvis Ridge tracks, establishing the age-progression pattern that anchors the modern hot-spot reference frame. The clean linear age progression along the Hawaiian-Emperor chain, from $0 Ma$ at Hawaii to $\sim 80 Ma$ at the Detroit Seamount, is the canonical evidence for the Morgan model.

Richards-Duncan-Ball 1989: flood basalts from plume heads

Mark Richards, Robert Duncan, and Robert Ball in 1989 ^{[Richards-Duncan-Ball 1989]} proposed that flood basalts (LIPs) are generated by the arrival of a plume head at the base of the lithosphere. The head delivers $1 0^{6}$ to $1 0^{7} k m^{3}$ of magma in $1 0^{5}$ to $1 0^{6}$ years, after which the plume tail sustains the long-lived hot-spot track. The Richards-Duncan-Ball model explained the temporal association of LIPs with the start of hot-spot tracks: the Deccan Traps at $66 Ma$ are followed by the Reunion hot-spot track in the Indian Ocean; the Ontong Java Plateau at $120 Ma$ is followed by the Louisville chain.

Campbell-Griffiths 1990: the plume-head/tail numerical model

Ian Campbell and Ross Griffiths in 1990 ^{[Campbell-Griffiths 1990]} published laboratory and numerical simulations of thermal plumes in a viscous fluid, showing that a plume initiating at a hot boundary layer develops a large bulbous head followed by a narrower tail, exactly as required by the Richards-Duncan-Ball model. The Campbell-Griffiths experiments produced quantitative predictions of head diameter ( $800$ to $1, 200 km$ ), temperature anomaly ( $200$ to $300 K$ ), and melt volume ( $1 0^{6}$ to $1 0^{7} k m^{3}$ ), all consistent with observed LIPs.

Courtillot 2003: the five-criteria plume test

Vincent Courtillot and colleagues in 2003 ^{[Courtillot 2003]} introduced a quantitative test for distinguishing primary plumes from secondary or doubtful ones. The five criteria, listed in the Formal Definition section, classify 7 to 9 of the catalogued hot spots as primary plumes of deep-mantle origin, with Hawaii, Iceland, Afar, Reunion, Louisville, Easter, Galapagos, Kerguelen, and Tristan satisfying most or all criteria. The Courtillot test provides a common language for the plume debate, allowing both sides to agree on which hot spots are uncontested.

Foulger 2005: the Plates vs Plumes critique

Gillian Foulger, Don Anderson, and colleagues in the 2005 GSA Special Paper Plates, Plumes, and Paradigms ^{[Foulger 2005]} articulated the strongest critique of the deep-plume hypothesis. The Foulger-Anderson Plate model proposes that most hot-spot volcanism is generated by shallow-mantle processes triggered by lithospheric extension, with no need for a deep thermal plume. The critique rests on three observations: many proposed plumes fail one or more Courtillot criteria, the depth resolution of seismic tomography in the lower mantle is too poor to image a CMB-anchored plume uniquely, and the geochemical signals (high $^{3} He /^{4} He$ ) can be explained by ancient shallow reservoirs without invoking deep isolation. The Plates vs Plumes debate remains geophysics' most contested question.

Wolfe 2009: Hawaiian plume tomography

Catherine Wolfe and colleagues in 2009 ^{[Wolfe 2009]} deployed ocean-bottom seismometers around Hawaii and imaged a mantle shear-wave velocity anomaly extending from the lithosphere through the transition zone and into the lower mantle to at least $1, 500 km$ depth. The seismic anomaly is consistent with a $200$ to $300 K$ thermal plume and provides the strongest direct tomographic evidence for a deep Hawaiian plume. The CMB connection at $2, 891 km$ remains below the resolution of the experiment, leaving the deep-plume hypothesis supported but not uniquely determined.

Synthesis. The plume hypothesis builds toward a unified mantle-convection picture in which Wilson's fixed hot spots, Morgan's plumes, the Richards-Duncan-Ball flood-basalt origin, and the Campbell-Griffiths head/tail geometry are layered, not alternative, descriptions. The foundational reason plumes matter for absolute plate reconstructions is that they are the only deep-mantle features with surface expression older than the preserved sea-floor; this is exactly the content of the hot-spot reference frame. The central insight is that putting these together, the same thermal-convection physics explains both the catastrophic short-lived flood basalts of the Deccan and Siberian Traps and the long-lived Hawaiian-Emperor chain, with the plume head and tail phases distinguished by flux rather than by mechanism. The pattern appears again in 27.01.01 plate tectonics, where absolute plate motions over $200 Myr$ are anchored by plume-track ages, and generalises to every tectonic plate that has crossed a plume; the bridge is the recognition that without mantle plumes, the deep Earth would be invisible to surface geology, and absolute plate motion through geologic time would be unrecoverable. Identifying the surface volcanic chain with a deep-mantle thermal anomaly is the structural move that gives plume geophysics its predictive power: from a single line of volcanoes, the entire absolute-motion history of a plate can be reconstructed.

Full proof set Master

Proposition (plume-head volume scaling). In the Campbell-Griffiths model, the plume-head volume $V_{h}$ scales as $Q^{3/5} κ^{- 2/5}$ , where $Q$ is the heat flux carried by the plume and $κ$ is the mantle thermal diffusivity. Consequently, doubling the plume heat flux increases the plume-head volume by a factor of $2^{3/5} \approx 1.52$ .

Proof. The plume head grows as the plume ascends through the mantle, entraining ambient mantle at a rate set by the buoyancy flux. The buoyancy flux $B$ is related to the heat flux by $B = α Q / c_{p}$ , where $α$ is the thermal expansivity and $c_{p}$ is the specific heat. The ascent time of the plume from the CMB ( $\sim 2, 900 km$ depth) to the surface is set by the Stokes velocity $U \sim B^{1/5} κ^{- 1/5}$ , giving $t_{ascent} \sim H / U$ , where $H$ is the mantle thickness.

During ascent, the plume head entrains ambient mantle at volumetric rate $d V / d t \sim B^{2/5} κ^{- 1/5}$ . The head volume at the surface is $V_{h} \sim (d V / d t) \cdot t_{ascent} \sim B^{3/5} κ^{- 2/5} H$ . Substituting $B = α Q / c_{p}$ gives the stated scaling, $V_{h} \propto Q^{3/5} κ^{- 2/5}$ .

For the empirical check: the Deccan Traps plume head delivered $\sim 1.5 \times 1 0^{6} k m^{3}$ of basaltic magma; the Siberian Traps delivered $\sim 4 \times 1 0^{6} k m^{3}$ . The ratio $4/1.5 \approx 2.7$ implies a heat-flux ratio of $2. 7^{5/3} \approx 5.1$ , consistent with the plume-head hypothesis if the Siberian plume was roughly five times more powerful than the Deccan, which is independently supported by the longer recurrence of Siberian-style events in the geologic record. $□$

Proposition (absolute plate velocity from a single hot-spot track). Let $x (t)$ be the position along a hot-spot track as a function of radiometric age $t$ , measured in a frame fixed to the overriding plate. Then the absolute velocity of the plate relative to the lower mantle, under the fixed-plume assumption, is

v_{plate} (t) = - \frac{d x}{d t}_{t} .

Proof. By hypothesis, the plume is fixed in the lower-mantle frame at position $x_{plume}$ . The active volcano at time $t$ sits directly above the plume, at position $x_{plume} + r$ where $r$ is the radial offset from the plume axis to the surface, with $∣ r ∣ = R_{Earth}$ . As the plate moves with absolute velocity $v_{plate}$ relative to the lower mantle, the volcano carried by the plate drifts to position $x (t)$ , leaving the plume beneath. The track left on the plate is the locus of past volcano positions, parameterized by age $t$ : $x (t) = x_{plume} + r + v_{plate} \cdot t$ . Differentiating, $d x / d t = v_{plate}$ , which is the velocity of the plate relative to the plume. Since the plume is fixed in the lower-mantle frame, $d x / d t$ equals the absolute plate velocity, with sign reversed because the track records where the plate was at past times, not where it is heading. $□$

Proposition (Courtillot primary-plume scoring). A plume candidate passing Courtillot criteria 1, 2, 3, and 5 (age progression, reference-frame stationarity, high $^{3} He /^{4} He$ , tomographic evidence) is classified as a primary plume of deep-mantle origin. The Hawaiian plume satisfies all four of these criteria and is therefore classified as primary.

Proof. (i) The Hawaiian-Emperor chain shows a monotonic age progression from $0$ to $\sim 80 Ma$ along a $6, 000 km$ line, satisfying criterion 1. (ii) When the Pacific Plate motion is reconstructed against multiple hot spots, the Hawaiian plume's reconstructed latitude has remained within $\sim 5^{\circ}$ of its present latitude over $80 Ma$ , satisfying criterion 2. (iii) Hawaiian basalts have $R / R_{A} \approx 16$ to $32$ , well above the MORB value of $8$ , satisfying criterion 3. (v) Seismic tomography beneath Hawaii (Wolfe et al. 2009 ^{[Wolfe 2009]}) images a slow seismic column extending from $\sim 400 km$ depth to at least $1, 500 km$ , with the lower-mantle extension inferred from global tomography models. Criterion 4 (surface heat flow) is partially satisfied but is the weakest criterion for oceanic hot spots because the insulating effect of the ocean makes direct measurement difficult. The plume therefore meets four of the five criteria strongly, qualifying as a primary plume. $□$

Connections Master

Plate tectonics and continental drift 27.01.01. The chapter survey introduces plate boundaries, relative plate motions, and the geometry of the lithosphere. This unit adds the missing piece: a deep-mantle-fixed reference frame against which absolute plate motion can be reconstructed. Every absolute plate-motion map published since 1971 uses hot-spot tracks as its anchor, so the plume hypothesis is methodologically inseparable from the global plate-kinematic framework established in the survey unit.
Megathrust earthquakes and the seismic cycle 27.03.04. Both units concern subduction-zone and mantle geodynamics. Where megathrust seismology measures the elastic-strain accumulation on the locked plate interface, plume volcanism measures the thermal-convection driving force that ultimately supplies the plate-motion energy. The convergence rate that the megathrust accumulates as slip deficit is set by the absolute plate velocity, which is independently measured by the hot-spot reference frame derived here; the two units therefore provide independent cross-checks on the same physical quantity.
Earth history and the geologic time scale 27.08.01. Large igneous provinces (Deccan, Siberian, CAMP, Ontong Java) correlate with mass extinctions and serve as age markers in the geologic timescale. The Deccan Traps at $66 Ma$ coincide with the K-Pg boundary, the Siberian Traps at $252 Ma$ with the Permian-Triassic boundary, and the CAMP province at $201 Ma$ with the Triassic-Jurassic boundary. Plume-head arrival is therefore not only a geophysical event but a biostratigraphic one, anchoring the timescale that the geologic-history unit relies on.
Stellar nucleosynthesis 28.02.05. The high $^{3} He /^{4} He$ ratio in plume basalts, central to Courtillot criterion 3, is interpreted as primordial helium trapped since Earth's accretion. The primordial helium itself was produced by Big Bang nucleosynthesis and by stellar helium-burning reactions catalogued in the stellar-nucleosynthesis unit. Connecting the deep-mantle helium reservoir to its nucleosynthetic origin closes a loop between mantle geochemistry and stellar physics, with plume basalts serving as the only accessible samples of the deep Earth's primordial inventory.

Historical & philosophical context Master

James Dwight Dana, sailing with the United States Exploring Expedition from 1838 to 1842, surveyed the Hawaiian volcanoes and noted in his 1890 monograph Characteristics of Volcanoes ^{[Dana 1890]} that the islands formed a progression consistent with a deep-seated magma source fixed beneath the moving crust. Dana's observation remained an isolated curiosity until the plate-tectonics revolution of the 1960s.

J. Tuzo Wilson, working at the University of Toronto in 1963, three years before Isacks, Sykes, and Oliver formalized plate tectonics, proposed the hot-spot hypothesis ^{[Wilson 1963]}. Wilson's paper in the Canadian Journal of Physics used the geometric progression of the Hawaiian Islands to argue for a fixed magma source beneath the moving Pacific Plate. The paper is the first identification of a deep-mantle-fixed reference frame for plate motions.

W. Jason Morgan, in a two-page 1971 paper in Nature ^{[Morgan 1971]}, generalized Wilson's intuition into the mantle plume hypothesis: thermal plumes rise from the lower mantle, fixed relative to one another, and their surface expression is hot-spot volcanism. Morgan identified about 20 global hot spots and proposed that they define a mantle-fixed reference frame. The Morgan paper is the founding document of plume geophysics.

The modern plume-head/tail model was crystallized in two papers: Richards-Duncan-Ball in 1989 ^{[Richards-Duncan-Ball 1989]} proposed that flood basalts originate from plume-head arrival, and Campbell-Griffiths in 1990 ^{[Campbell-Griffiths 1990]} confirmed the head/tail geometry with laboratory and numerical experiments. Vincent Courtillot in 2003 ^{[Courtillot 2003]} formalized the five-criteria plume test, providing a quantitative framework for distinguishing deep plumes from shallow-mantle volcanism.

Gillian Foulger, Don Anderson, and colleagues in the 2005 GSA Special Paper Plates, Plumes, and Paradigms ^{[Foulger 2005]} articulated the strongest critique of the plume hypothesis, arguing that shallow-mantle processes and lithospheric extension explain many proposed plumes without invoking deep-mantle convection. The Plates vs Plumes debate remains geophysics' most contested question, with the Hawaiian plume the canonical case for the plume school and the Columbia River Basalts the canonical case for the Plate school.

Bibliography Master

@article{Wilson1963,
  author = {Wilson, J. Tuzo},
  title = {A possible origin of the {Hawaiian} {Islands}},
  journal = {Canadian Journal of Physics},
  volume = {41},
  number = {5},
  pages = {993--999},
  year = {1963},
}

@article{Morgan1971,
  author = {Morgan, W. Jason},
  title = {Convection plumes in the lower mantle},
  journal = {Nature},
  volume = {230},
  pages = {42--43},
  year = {1971},
}

@article{Crough1978,
  author = {Crough, S. Thomas},
  title = {Thermal origin of mid-plate hot-spot swells},
  journal = {Geophysical Journal of the Royal Astronomical Society},
  volume = {55},
  number = {2},
  pages = {451--469},
  year = {1978},
}

@article{Duncan1981,
  author = {Duncan, Robert A.},
  title = {Hotspots in the southern oceans---an absolute frame of reference for motion of the {Gondwana} continents},
  journal = {Tectonophysics},
  volume = {74},
  number = {1-2},
  pages = {29--42},
  year = {1981},
}

@article{RichardsDuncanBall1989,
  author = {Richards, Mark A. and Duncan, Robert A. and Ball, Valentine E.},
  title = {Flood basalts and hot-spot tracks: plume heads and tails},
  journal = {Science},
  volume = {246},
  number = {4926},
  pages = {103--107},
  year = {1989},
}

@article{CampbellGriffiths1990,
  author = {Campbell, Ian H. and Griffiths, Ross W.},
  title = {Implications of mantle plume structure for the evolution of flood basalts},
  journal = {Earth and Planetary Science Letters},
  volume = {99},
  number = {1-2},
  pages = {79--93},
  year = {1990},
}

@article{Courtillot2003,
  author = {Courtillot, Vincent and Davaille, Anne and Besse, Jean and Stock, Joann},
  title = {Three distinct types of hotspots in the {Earth's} mantle},
  journal = {Earth and Planetary Science Letters},
  volume = {205},
  number = {3-4},
  pages = {295--308},
  year = {2003},
}

@book{Foulger2005,
  editor = {Foulger, Gillian R. and Natland, James H. and Presnall, Dean C. and Anderson, Don L.},
  title = {Plates, Plumes, and Paradigms},
  series = {GSA Special Paper},
  number = {388},
  publisher = {Geological Society of America},
  year = {2005},
}

@article{Wolfe2009,
  author = {Wolfe, Catherine J. and Solomon, Sean C. and Laske, Gabi and Collins, John A. and Bye, Rachel S. and Orcutt, John A. and Bercovici, David and Hauri, Erik H.},
  title = {Mantle shear-wave velocity structure beneath the {Hawaiian} hot spot},
  journal = {Science},
  volume = {326},
  number = {5958},
  pages = {1388--1390},
  year = {2009},
}

@book{Dana1890,
  author = {Dana, James Dwight},
  title = {Characteristics of Volcanoes, with Contributions of Facts to the Principles of Geology},
  publisher = {Dodd, Mead and Company},
  address = {New York},
  year = {1890},
}

@book{Condie2015,
  author = {Condie, Kent C.},
  title = {Earth as an Evolving Planetary System},
  edition = {3},
  publisher = {Academic Press},
  year = {2015},
}

@book{Fortey2004,
  author = {Fortey, Richard},
  title = {Earth: An Intimate History},
  publisher = {HarperCollins},
  year = {2004},
}

Prerequisites

27.01.01

Tier anchors

beginner: Richard Fortey, Earth: An Intimate History (2004); Tarbuck and Lutgens, Earth Science (15e), Ch. 2 hot-spot section
intermediate: Condie, Earth as an Evolving Planetary System (3e); Storey, Mantle Convection and Plate Tectonics in 'Dynamic Earth' (Cambridge)
master: Wilson 1963 Can. J. Phys. 41:993; Morgan 1971 Nature 230:42; Richards-Duncan-Ball 1989; Campbell-Griffiths 1990; Courtillot 2003; Foulger 2005

References

Wilson, J. T. — A possible origin of the Hawaiian Islands, Canadian Journal of Physics 41, 993-999 (1963) · Full text; the hot-spot hypothesis
Morgan, W. J. — Convection plumes in the lower mantle, Nature 230, 42-43 (1971) · Full text; the founding plume paper
Richards, M. A., Duncan, R. A. and Ball, V. E. — Flood basalts and hot-spot tracks: plume heads and tails, Science 246, 103-107 (1989) · Full text; plume-head/tail model
Campbell, I. H. and Griffiths, R. W. — Implications of mantle plume structure for the evolution of flood basalts, Earth and Planetary Science Letters 99, 79-93 (1990) · Full text; plume-head/tail numerical model
Courtillot, V., Davaille, A., Besse, J. and Stock, J. — Three distinct types of hotspots in the Earth's mantle, Earth and Planetary Science Letters 205, 295-308 (2003) · Full text; 5-criteria plume test
Foulger, G. R., Natland, J. H., Presnall, D. C. and Anderson, D. L. (eds.) — Plates, Plumes, and Paradigms, GSA Special Paper 388 (2005) · Edited volume; the Plates vs Plumes critique
Wolfe, C. J., Solomon, S. C., Laske, G., Collins, J. A., Bye, R. S., Orcutt, J. A., Bercovici, D. and Hauri, E. H. — Mantle shear-wave velocity structure beneath the Hawaiian hot spot, Science 326, 1388-1390 (2009) · Full text; Hawaiian plume tomography
Dana, J. D. — Characteristics of Volcanoes, with Contributions of Facts to the Principles of Geology (1890) · Dana on the Hawaiian deep-seated source

Estimated time

beginner: 25m
intermediate: 50m
master: 85m