Nature | Vol 586 | 15 October 2020 | 395Article
Building cratonic keels in Precambrian plate
tectonics
A. L. Perchuk1,2 ✉, T. V. Gerya^3 , V. S. Zakharov^1 & W. L. Griffin^4The ancient cores of continents (cratons) are underlain by mantle keels—volumes of
melt-depleted, mechanically resistant, buoyant and diamondiferous mantle up to
350 kilometres thick, which have remained isolated from the hotter and denser convecting
mantle for more than two billion years. Mantle keels formed only in the Early Earth
(approximately 1.5 to 3.5 billion years ago in the Precambrian eon); they have no
modern analogues^1 –^4. Many keels show layering in terms of degree of melt depletion^5 –^7.
The origin of such layered lithosphere remains unknown and may be indicative of a
global tectonics mode (plate rather than plume tectonics) operating in the Early
Earth. Here we investigate the possible origin of mantle keels using models of oceanic
subduction followed by arc-continent collision at increased mantle temperatures
(150–250 degrees Celsius higher than the present-day values). We demonstrate that
after Archaean plate tectonics began, the hot, ductile, positively buoyant, melt-depleted
sublithospheric mantle layer located under subducting oceanic plates was unable to
subduct together with the slab. The moving slab left behind craton-scale
emplacements of viscous protokeel beneath adjacent continental domains. Estimates
of the thickness of this sublithospheric depleted mantle show that this mechanism
was efficient at the time of the major statistical maxima of cratonic lithosphere ages.
Subsequent conductive cooling of these protokeels would produce mantle keels with
their low modern temperatures, which are suitable for diamond formation.
Precambrian subduction of oceanic plates with highly depleted mantle is thus a
prerequisite for the formation of thick layered lithosphere under the continents,
which permitted their longevity and survival in subsequent plate tectonic processes.The paradigm of plate tectonics is a powerful tool for investigation
of key geological processes and related hazards in the modern Earth.
However, it remains highly uncertain how far back in time towards
the hot early Earth^8 this paradigm can be extended^6 ,^9 –^11. A large obsta-
cle to understanding the evolution of the Earth since the Archaean
(4–2.5 billion years ago, Ga) is the poor geological record preserved
exclusively in the crust of cratons^12 and in their thick depleted mantle
keels^5 ,^13. The origin of these important geological formations remains
unknown in terms of tectonic regime and the scope of any link between
crust-forming and mantle-lithosphere-forming processes^11.
The mantle keels are the diamond treasury of the planet; they were
formed only in the Archaean (before 2.5 Ga) and in the Early Proterozoic
(1.6–2.5 Ga), and have no modern analogues^1 –^4. They are fundamen-
tally different from other types of continental lithospheric mantle in a
number of specific ways, such as their great thickness (hence the name
‘keel’), low temperature and low density^3. Many keels show layering in
terms of degree of melt depletion^5 –^7. The older the keels are the colder
and less dense they are^14 ,^15. The low density of the keels is attributed
to their specific chemical composition^13 ,^14 , which indicates very high
degrees of melt depletion that are not realized in the modern Earth but
potentially could have been produced in Archaean spreading ridges^1.
The mantle keels should be mechanically resistant^1 , precluding their
destruction by convecting mantle, which would explain their longevity.
The formation of the mantle keels is generally discussed in terms of
two contrasting end-member models. One model involves high-degree
melting in rising mantle diapirs/plumes or mantle overturns^1 ,^6 ,^13 ,^16 –^18 ,
that is, mantle keels represent the melt-depleted, dry and low-density
residues of melting^1 ,^6 ,^13. This model is very attractive because it pro-
vides reasonable explanations of various observations: (1) low den-
sity (high degree of melt depletion) and thus buoyancy of the keels^1 ,^6 ;
(2) positive correlation of keel thickness with age (effect of potential
mantle temperature)^14 ,^15 ; and (3) dry peridotite xenoliths that indicate
sufficient viscosity contrasting with the underlying asthenosphere^2.
The model predicts that the mantle keels should be relatively more
dense at the base and less dense at the top of the melting column
because of the increased melt extraction towards the surface during
decompression-melting processes. However, chemical stratification
of the mantle keels corresponds only poorly to what is predicted by the
model^3 ,^19. The second model postulates that mantle keels were made
by stacking slabs of subducted oceanic crust and mantle consistinghttps://doi.org/10.1038/s41586-020-2806-7
Received: 22 December 2019
Accepted: 26 August 2020
Published online: 14 October 2020
Check for updates(^1) Geological Faculty, Lomonosov Moscow State University, Moscow, Russia. (^2) Korzhinskii Institute of Experimental Mineralogy, Russian Academy of Sciences, Chernogolovka, Russia. (^3) Swiss
Federal Institute of Technology Zurich, Department of Earth Sciences, Zurich, Switzerland.^4 Australian Research Council Centre of Excellence for Core to Crust Fluid Systems/GEMOC,
Macquarie University, Sydney, New South Wales, Australia. ✉e-mail: [email protected]