generation, based on diamond stability and experimental petrology, is ~150 km 6,7 but some 26 have suggested far deeper generation depths (400-600 km 8 or even > 660-1700 km 9,10 ) for 27 kimberlite. Here we put all these results into a wider perspective by demonstrating that most 28 Large igneous provinces (LIPs) consist dominantly of basaltic rock erupted relatively 31 rapidly (1-5 Myr) over great areas (1-10 x 10 6 km 2 ) 12 . Earlier work has shown that most LIPs 32 of the past 300 Myr, rotated back to their eruption sites, and active deep-plume sourced 33 hotspots at the Earth's surface (Fig. 1), project radially down to lie on narrow stable PGZs on 34 the CMB at the edge of the hot and dense large low shear wave velocity provinces 35 (LLSVPs 13 ) of the deep mantle 11,14-19 , thus demonstrating long-term stability of LLSVPs. The 36 1% slow velocity contour in the lowermost layer of the SMEAN tomography model 20 is a 37 reasonable proxy for the PGZs because most reconstructed LIP eruption sites and steep 38 horizontal gradients in shear-wave anomalies in the SMEAN model fall close to that 39 contour 14 . In Figure 1 we show twelve hotspots found by seismic tomography 18 to be sourced 40 by deep plumes. Some other hotspots, which have also been claimed to be sourced from deep 41 plumes using other selection criteria (e.g. Tristan, Reunion, Afar and Hawaii), are not shown 42 on our map, but they too plot close to vertically above the PGZs 14, 16 . 43To find out whether kimberlites show an association with PGZs similar to that shown by 44LIPs and hotspots, we used plate reconstructions 21,22 to rotate kimberlites that are younger 45 than the initial assembly of Pangea (~320 Ma), to their original eruption sites. We find that 46 eighty percent of kimberlites (1112 of 1395) of the past 320 Myr were erupted when their 47 eruption sites lay above a half-width of 15 o on either side of the 1% slow contour of SMEAN 48 in the lowermost mantle beneath Africa (Fig. 1). On average, this dominant part of the 49 kimberlite population plots at a distance of 7 ± 5 o from that contour (online supplementary 50 material Table S1). The most anomalous post-320 Ma kimberlites (17%) are in the Slave 51Province of Canada (Late Cretaceous/Early Tertiary kimberlites 2 ), which was close to a 52 tectonically active continental margin at the times of their eruption. 53 A remarkable pattern is observed when we plot kimberlites on our series of plate 54 reconstructions. At practically all times, eruption sites plot close to the African PGZ ( Fig. 2; 55 S2-S5). For the past 320 Myr, Gondwana with Africa at its heart, has drifted slowly 56 northward over the African PGZ (online supplementary material Figs. S2-S5), and this readily 57 explains the dominance of African (Gondwana) kimberlites in the global record, if, as we 58 suggest, their origin relates to heat from deep plumes. Globally, kimberlite activity peaked 59 between 70 and 120 Ma (Fig. S1), corresponding to the time of formation of some of the most 60 economically viable diamond...

No abstract

Earth's residual geoid is dominated by a degree-2 mode, with elevated regions above large low shear-wave velocity provinces on the core-mantle boundary beneath Africa and the Pacific. The edges of these deep mantle bodies, when projected radially to the Earth's surface, correlate with the reconstructed positions of large igneous provinces and kimberlites since Pangea formed about 320 million years ago. Using this surface-to-core-mantle boundary correlation to locate continents in longitude and a novel iterative approach for defining a paleomagnetic reference frame corrected for true polar wander, we have developed a model for absolute plate motion back to earliest Paleozoic time (540 Ma). For the Paleozoic, we have identified six phases of slow, oscillatory true polar wander during which the Earth's axis of minimum moment of inertia was similar to that of Mesozoic times. The rates of Paleozoic true polar wander (<1°/My) are compatible with those in the Mesozoic, but absolute plate velocities are, on average, twice as high. Our reconstructions generate geologically plausible scenarios, with large igneous provinces and kimberlites sourced from the margins of the large low shear-wave velocity provinces, as in Mesozoic and Cenozoic times. This absolute kinematic model suggests that a degree-2 convection mode within the Earth's mantle may have operated throughout the entire Phanerozoic.plate reconstructions | thermochemical piles T wo equatorial, antipodal, large low shear-wave velocity provinces ( Fig. 1) in the lowermost mantle (1) beneath Africa (termed Tuzo) (2) and the Pacific Ocean (Jason) are prominent in all shear-wave tomographic models (3-7) and have been argued to be related to a dominant degree-2 pattern of mantle convection that has been stable for long times (3). Most reconstructed large igneous provinces and kimberlites over the past 300 My have erupted directly above the margins of Tuzo and Jason, which we term the plume generation zones (1, 2, 5). This remarkable correlation suggests that the two deep mantle structures have been stable for at least 300 My. Stability before Pangea (before 320 Ma) is difficult to test with plate reconstructions because the paleogeography, the longitudinal positions of continents, and the estimates of true polar wander are uncertain (8). It is similarly challenging to reproduce such long-term stability in numerical models (9). However, if the correlation between the eruption sites of large igneous provinces, kimberlites, and the plume generation zones observed for the past 300 Ma has been maintained over the entire Phanerozoic (0-540 Ma), it can provide a crucial constraint for defining the longitudinal positions of continental blocks during Paleozoic time (250-540 Ma).Here we show that a geologically reasonable kinematic model that reconstructs continents in longitude in such a way that large igneous provinces and kimberlites are positioned above the plume generation zones at the times of their formation ( Fig. 2A and SI Appendix, Fig. S2) can be successfully defined f...

Using a recent compilation of African alkaline igneous rocks and carbonatites, we show that nearly 90% (28 of 32 occurrences) of nepheline syenite gneisses and deformed carbonatites are concentrated within known or inferred Proterozoic suture zones. Given the wellestablished intracontinental rift setting for these rocks and the likely continental collisional setting for their subsequent deformation, we suggest that deformed alkaline rocks and carbonatites (DARCs) represent the products of two well-defined parts of the Wilson Cycle. DARCs mark the places where vanished oceans have opened and then closed. We further postulate that DARCs taken into the mantle lithosphere to ca. 100 km depths at collision could provide source material for later alkaline magmatic activity. This could account for the observation of recurrent alkaline magmatic activity over hundreds of m.y. in provinces such as that of southern Malawi.

No abstract