Archean komatiite volcanism controlled by the evolution of early continents
- aCentre for Exploration Targeting, Australian Research Council Centre of Excellence for Core to Crust Fluid Systems, School of Earth and Environment, University of Western Australia, Perth, WA 6009, Australia;
- bDepartment of Applied Geology, Curtin University, Bentley, WA 6102, Australia;
- cGeological Survey of Western Australia, Department of Mines and Petroleum, East Perth, WA 6004, Australia;
- dKey Centre for the Geochemical Evolution and Metallogeny of Continents, Australian Research Council Centre of Excellence for Core to Crust Fluid Systems, Macquarie University, North Ryde NSW 2109, Australia; and
- eEarth Science and Resource Engineering, Commonwealth Scientific and Industrial Research Organization (CSIRO), Kensington, Perth, WA 6151, Australia
See allHide authors and affiliations
Edited by Norman H. Sleep, Stanford University, Stanford, CA, and approved April 14, 2014 (received for review January 7, 2014)

Significance
Komatiites are rare, ultra-high-temperature (∼1,600 °C) lavas that were erupted in large volumes 3.5–1.5 bya but only very rarely since. They are the signature rock type of a hotter early Earth. However, the hottest, most extensive komatiites have a very restricted distribution in particular linear belts within preserved Archean crust. This study used a combination of different radiogenic isotopes to map the boundaries of Archean microcontinents in space and time, identifying the microplates that form the building blocks of Precambrian cratons. Isotopic mapping demonstrates that the major komatiite belts are located along these crustal boundaries. Subsequently, the evolution of the early continents controlled the location and extent of major volcanic events, crustal heat flow, and major ore deposit provinces.
Abstract
The generation and evolution of Earth’s continental crust has played a fundamental role in the development of the planet. Its formation modified the composition of the mantle, contributed to the establishment of the atmosphere, and led to the creation of ecological niches important for early life. Here we show that in the Archean, the formation and stabilization of continents also controlled the location, geochemistry, and volcanology of the hottest preserved lavas on Earth: komatiites. These magmas typically represent 50–30% partial melting of the mantle and subsequently record important information on the thermal and chemical evolution of the Archean–Proterozoic Earth. As a result, it is vital to constrain and understand the processes that govern their localization and emplacement. Here, we combined Lu-Hf isotopes and U-Pb geochronology to map the four-dimensional evolution of the Yilgarn Craton, Western Australia, and reveal the progressive development of an Archean microcontinent. Our results show that in the early Earth, relatively small crustal blocks, analogous to modern microplates, progressively amalgamated to form larger continental masses, and eventually the first cratons. This cratonization process drove the hottest and most voluminous komatiite eruptions to the edge of established continental blocks. The dynamic evolution of the early continents thus directly influenced the addition of deep mantle material to the Archean crust, oceans, and atmosphere, while also providing a fundamental control on the distribution of major magmatic ore deposits.
Volcanism on Earth is the dynamic surface expression of our planet’s thermal cycle, with heat created from radioactive decay and lost through mantle convection (1). In the Archean eon (>2.5 bya), Earth’s heat flux was significantly higher than that observed today (1, 2) due to the combined effects of a more radioactive mantle (1, 3) and residual heat from planetary accretion (4). This resulted in the eruption of komatiites: ultra-high temperature, low-viscosity lavas with MgO >18% and eruption temperatures >1,600 °C (5) formed from mantle plumes (1, 2, 6). These rare, ancient magmas are dominantly restricted to the early history of the planet (3.5–1.5 Ga; ref. 7) and represent the remnants of huge volcanic flow fields (8) consisting of the hottest lavas preserved on Earth (5, 9, 10). These now-extinct volcanic systems and flow complexes had the potential to cover significant portions of the early continents, and were likely analogous to large igneous provinces in size and magma volume (11, 12). Komatiites are vital to our understanding of Earth’s thermal evolution (1⇓–3, 7, 13⇓⇓–16), and represent a window into the dynamic secular development of the mantle throughout the early history of our planet (5). Subsequently, understanding the physical and chemical processes that govern their localization, volcanology, and geochemistry is vital in deciphering this information.
In the Yilgarn Craton of Western Australia (Fig. 1), two major pulses of komatiite activity occurred at ∼2.9 Ga (southern Youanmi Terrane; refs. 17⇓–19) and 2.7 Ga (Kalgoorlie Terrane, Eastern Goldfields Superterrane; refs. 5, 10). These represent two separate plume events that impinged onto preexisting continental crust (20⇓⇓–23), with the resulting magmas heterogeneously distributed across the craton (8, 10, 17⇓–19, 23, 24). In this study, we provide the first evidence of a fundamental relationship between the spatiotemporal variation in komatiite abundance, geochemistry, and volcanology and the evolution of an Archean microcontinent, reflected in the changing isotopic composition of the crust.
Map of the Archean Yilgarn Craton showing the basic granite-greenstone bedrock geology and location of the ∼2.9 and 2.7 Ga komatiite localities. Individual terranes/domains (39, 40) are labeled. Greenstone belts are labeled as follows: MD, Marda–Diemals; SC, Southern Cross; FO, Forrestania; LJ, Lake Johnston; RAV, Ravensthorpe; AW, Agnew–Wiluna; and KAL, Kalgoorlie/Kambalda. Komatiite localities are from Barnes and Fiorentini (10) (Table S4).
We used Lu-Hf and U-Pb isotopic techniques on multiple magmatic and inherited zircon populations from granitoid rocks and felsic volcanic units, which represent the exposed Archean crust of the Yilgarn Craton. All zircon grains were dated using the sensitive high-resolution ion microprobe (SHRIMP), before in situ laser ablation inductively coupled plasma mass-spectrometry (LA-ICP-MS) analysis for Lu-Hf isotopes. The Lu-Hf isotopic data are expressed as εHf, which denotes the derivation of the 176Hf/177Hf ratio of the sample from the contemporaneous ratio of the chondritic uniform reservoir (CHUR), multiplied by 104. The term “juvenile” refers to crustal material that plots on or close to the depleted mantle evolution line, suggesting derivation from a depleted mantle source. In contrast, “reworked” refers to the remobilization of preexisting crust by partial melting and/or erosion and sedimentation (25, 26). Complete sample information, methodology, and geochemical datasets (U-Pb, Lu-Hf, and komatiite) are available in the Supporting Information, Figs. S1–S3, and Tables S1–S4.
The Lu-Hf data are displayed as a series of time-slice contour maps, which show “snapshots” of the changing source and age of the crust at 3,050–2,820; 2,820–2,720; and 2,720–2,600 Ma [Figs. 2–4; intervals based on the work of Mole et al. (21); full Hf dataset displayed in Fig. S4]. In these maps, point data representing the median εHf value of granites and felsic volcanics are plotted as contour maps that show the spatial extent of “blocks” of specific Lu-Hf isotopic character and their evolution through time. This method is based on previous isotopic mapping of the Yilgarn Craton using the analogous Sm-Nd system (27). Importantly, the Lu-Hf data presented here replicate the features of the Sm-Nd work (27, 28), with the added ability to look further back in time due to the in situ analysis of abundant inherited zircons (21).
Lu-Hf (εHf) map of the Yilgarn Craton at 3,050–2,820 Ma. (A) Hf isotope map with the location of sample sites and komatiite localities. (B) Interpretive map of the area, showing the individual crustal blocks identified from the Hf isotope map and corresponding probability density plots. The blue curve represents the median εHf for discrete temporal groups (ng), whereas the red curve represents all of the individual grain analyses (na). Dark gray polygons shown in the background of all maps represent supracrustal belts (Fig. 1).
Lu-Hf (εHf) map of the Yilgarn Craton at 2,820–2,720 Ma. (A) Hf isotope map with the location of sample sites and komatiite localities. (B) Interpretive map of the area, showing the individual crustal blocks identified from the Hf isotope map and corresponding probability density plots. The blue curve represents the median εHf for discrete temporal groups (ng), whereas the red curve represents all of the individual grain analyses (na). Dark gray polygons shown in the background of all maps represent supracrustal belts (Fig. 1).
Lu-Hf (εHf) map of the Yilgarn Craton at 2,720–2,600 Ma. (A) Hf isotope map with the location of sample sites and komatiite localities. (B) Interpretive map of the area, showing the individual crustal blocks identified from the Hf isotope map and corresponding probability density plots. The blue curve represents the median εHf for discrete temporal groups (ng), whereas the red curve represents all of the individual grain analyses (na). Dark gray polygons shown in the background of all maps represent supracrustal belts (Fig. 1).
The variable isotopic signatures of the crust (Figs. 2–4) can be interpreted as proxies for lithospheric thickness (Figs. 5 and 6; ref. 29), where young, juvenile εHf values (εHf > 0) indicate relatively thin lithosphere and old, reworked values (εHf < 0) reflect thicker lithosphere (29, 30); a pattern observed in the modern-day western United States (29⇓⇓–32). Here, this information is combined to document the four-dimensional lithospheric architecture of the Yilgarn Craton and development of an Archean microcontinent.
Isotopic cross-section and interpreted lithospheric architecture during the emplacement of ∼2.9 Ga komatiites in the southern Youanmi Terrane. (A) εHf map showing the isotopic architecture at the time of the ∼2.9 Ga plume emplacement, with the approximate extent of the plume head (red) and tail (yellow) shown for scale; (B) isotopic cross-section (A–A′) documenting the changing εHf of the crust from east to west (circles represent median; squares represent individual analyses) together with the occurrence and MgO content (Table S4) of ultramafic–mafic magmatism; and (C) interpreted lithospheric architecture based on the changing isotopic properties of the crust. The white ellipses represent the types of magma available in a particular area and the dashed lines show the approximate limits of their source regions. TiO2 vs. Al2O3 data (17, 19) are shown for komatiites of the relevant greenstone belts, demonstrating the progressive eastward homogenization of Barberton-type melts. The location of the continent core and continent edge are shown based on the Lu-Hf data. The eruption of komatiite would likely have been facilitated by plume-related extension at this interface. Approximate thickness values for developed Archean lithosphere (∼250–150 km) were taken from Boyd et al. (41), Artemieva and Mooney (42), and Begg et al. (43). The approximate scale of the plume head (∼1,600 km), tail (200–100 km), and thickness (∼150–100 km) were taken from Campbell et al. (15). Note that the plume-tail material moves above the plume head, despite impacting the lithosphere later, as it is hotter, more buoyant, and subsequently emplaced at higher flux (15).
Isotopic cross-sections and interpreted lithospheric architecture during the emplacement of the ∼2.7 Ga komatiites in the Eastern Goldfields (Kalgoorlie Terrane). (A) εHf map showing the isotopic architecture at the time of the ∼2.7 Ga plume emplacement, with the approximate extent of the plume head (red) and tail (yellow) shown for scale; (B) isotopic cross-section (B–B′) documenting the changing εHf of the crust from east to west (circles represent median; squares represent individual analyses) together with the occurrence and MgO content (Table S4) of ultramafic–mafic magmatism; and (C) interpreted lithospheric architecture based on the changing isotopic properties of the crust. The white ellipses represent the types of magma available in a particular area and the dashed lines show the approximate limits of their source regions. TiO2 vs. Al2O3 data (9) are shown for komatiites of the Eastern Goldfields, demonstrating the eastward dilution and removal of Barberton-type melts. The location of the continent core and continent edge are shown based on the Lu-Hf dataset. The eruption of komatiite would likely have been facilitated by plume-related extension at this interface. The dashed red lines shown in C account for the potential variation in lithospheric architecture within the Lake Johnston block based on the Lu-Hf data. External data used to construct these diagrams are the same as for Fig. 5. Note that the plume-tail material moves above the plume head, despite impacting the lithosphere later, as it is hotter, more buoyant, and subsequently emplaced at higher flux (15).
Results
The first time slice (T1 − 3,050–2,820 Ma; Fig. 2) shows the lithospheric architecture at the time of ∼2.9 Ga komatiite emplacement in the southern Youanmi Terrane. The Lu-Hf mapping identifies three lithospheric blocks: Marda, Hyden, and Lake Johnston. The Marda and Hyden blocks dominantly comprise reworked older crust, with εHf −6.0 and −4.0, respectively. In contrast, the Lake Johnston block comprises younger, more juvenile material, with εHf +2.0. This protocratonic lithospheric architecture exerts a first-order control on the localization of the ∼2.9 Ga high-MgO komatiites of the Forrestania (17) and Lake Johnston (10, 19, 33) greenstone belts. These komatiites occur in the juvenile Lake Johnston block, primarily along the margin of the older Hyden block (Fig. 2). Within the Marda block to the north, komatiites are largely absent and the greenstone stratigraphy is almost exclusively comprised of basalt (24). No known volcanic units from this time are preserved in the Hyden block.
The second time slice (T2 − 2,820–2,720 Ma; Fig. 3) images the lithospheric architecture of the craton at a time when no significant evidence of komatiite magmatism is recorded. Six crustal blocks can be identified: Barlee, Marda, Hyden, Corrigin, Lake Johnston, and Eastern Goldfields. The Barlee block shows a bimodal εHf distribution, with peaks at 0 and +3.0. In relation to the first time slice, the Hyden block has extended north and merged with the Marda block, although its western extent appears to have been rejuvenated by mantle input (Corrigin block – εHf ∼0). The result is a narrow block of old, reworked crust with εHf peaks at −4.3 and −2.3. The Marda block remains old and unradiogenic, with a major εHf peak at −4.0. The Lake Johnston block has been significantly reworked since 3,050–2,820 Ma, as demonstrated by the lower εHf at −0.5 (Fig. 3). The Eastern Goldfields is now identifiable as a crustal block, and is more juvenile than the blocks to the west, displaying a bimodal εHf distribution with peaks at +2.0 and +4.0. Overall, the second time slice shows reworking throughout the core area of the West Yilgarn and addition of juvenile material at the northern, western, and eastern edges.
The third time slice (T3 − 2,720–2,600 Ma; Fig. 4) represents the lithospheric architecture at the time of voluminous ∼2.7 Ga komatiite volcanism in the Kalgoorlie Terrane of the Eastern Goldfields Superterrane (10, 33). Two major lithospheric blocks are present at this time: the Eastern Goldfields and the West Yilgarn (Fig. 4). The Eastern Goldfields block comprises a bimodal εHf distribution, with the majority of material at +2.0, and a notable minor peak at −2.0. The West Yilgarn is the result of the progressive cratonization of the individual blocks identified in time slices T1 and T2. It has a εHf distribution with peaks at −2.4 (major) and +1.8, −5.5, and −8.0 (minor). These variable sources reflect the complex history of the West Yilgarn as well as the intracratonic signature of the individual blocks from which it derives (Fig. 4). Overall, it appears that the majority of the West Yilgarn formed from a source ∼800 Ma (Marda) to 200 Ma (Lake Johnston) older than that of the Eastern Goldfields (Fig. S4).
Discussion
In the Yilgarn Craton, high-flux, thick channelized komatiites with average whole-rock MgO contents >30% only occur in greenstone belts located at the interface between juvenile and reworked crustal domains (Figs. 2 and 4). Preexisting lithospheric weaknesses at these craton margins (10, 34, 35) led to crustal attenuation, major intracontinental rifts, and high-flux magma transport through associated translithospheric conduits (10, 34). Ultramafic magmas rapidly ascended from mantle to surface without significant ponding or differentiation in the lithosphere (Figs. 5 and 6) (34, 35), and formed giant, high-flux komatiite flow fields that contain large nickel-sulfide ore reserves (5, 33). At ∼2.9 Ga in the juvenile Lake Johnston block, channelized high-MgO komatiites erupted to form thick olivine cumulate bodies that are now preserved in the Forrestania (17), Lake Johnston (10, 19, 33), and Ravensthorpe (36) greenstone belts. Their setting is consistent with the occurrence of a paleocraton margin at the Hyden–Lake Johnston block boundary at ∼2.9 Ga (Figs. 2 and 5). Between ∼2.9 and 2.7 Ga, the focus of hot and voluminous komatiite magmatism shifted to the east toward the Kalgoorlie Terrane of the Eastern Goldfields (Fig. 6), following the margin of the growing continent.
Conversely, plume-derived magmas that erupted in “intracontinent” settings (Figs. 5 and 6) formed abundant basalts and only low-flux, thin komatiites with average MgO values typically <30% (10, 33), as demonstrated by the greenstone belts of the Marda block at ∼2.9 Ga (24) and in the Kurnalpi Terrane at ∼2.7 Ga (10). In this setting, magmas rose through thick reworked lithosphere (Marda) or thinner juvenile (Kurnalpi) crust that was not adjacent to an older, thicker crustal block at the time of magmatism (Figs. 2 and 4). Consequently, eruptions were not sufficiently voluminous or continuous to form the giant komatiite flow fields and related nickel mineralization that is associated with craton margins.
The komatiites of the Southern Cross greenstone belt (18) form an intermediate group between the intracontinent (low-MgO, unchannelized) and the continent edge (high-MgO, channelized) type komatiites. These magmas have high-MgO contents (18), but are unchannelized, cumulate-poor, and unmineralized (18). We suggest that, due to the unique location of the Southern Cross greenstone belt between the reworked Hyden and Marda blocks (Fig. 2), magmas were focused enough to form high-MgO komatiites, but not to the extent needed for high-flux, continuous eruptions.
The heterogeneous nature of komatiite magmatism at ∼2.9 and 2.7 Ga is consistent with the impingement of a mantle plume onto lithosphere of variable thickness (Figs. 5 and 6). The isotopic architecture constrained in this study indicates that the ∼2.9 and 2.7 Ga komatiites were emplaced in similar geodynamic settings, at the margins of thick crustal domains (Figs. 2–6). Consequently, the melting dynamics of the underlying mantle and the petrogenetic signature of the komatiites would also be expected to be similar. However, although both Munro- and Barberton-type komatiites were erupted in the Lake Johnston block at ∼2.9 Ga (17⇓–19, 36), only Munro-type komatiite magmatism is recorded in the Kalgoorlie Terrane at ∼2.7 Ga (10, 19, 33).
Barberton-type komatiites contain notably depleted aluminum concentrations (Al2O3/TiO2 ∼10) in relation to Munro-type magmas (Al2O3/TiO2 ∼20). These compositional differences reflect the conditions under which the melts separated from their plume sources (2, 5). Barberton-type komatiites formed in the garnet stability zone, at a depth of ∼450–300 km (37); whereas, Munro-type komatiites segregated from their mantle sources at <300 km depth, although high-percentage partial melts (>30%) only formed at <150 km depth (2, 5). A possible explanation for the occurrence of Barberton-type komatiites in the ∼2.9 Ga sequences is the secular cooling of the Earth; melt generation at >300 km is only possible in the hotter, older mantle (13). This hypothesis is supported by the well-constrained global decline of Barberton-type komatiite magmatism from 3.5 to 2.7 Ga (2, 5). However, it is unlikely that sufficient global cooling occurred between ∼2.9 and 2.7 Ga to affect the source depth of the Yilgarn komatiites (1, 2).
An alternative scenario considers the spatiotemporal variability in komatiite type as a consequence of petrogenetic filtering by thick lithosphere. The relative architecture between the ∼2.9 Ga Hyden−Lake Johnston and ∼2.7 Ga West Yilgarn−Eastern Goldfields blocks is fundamentally similar (Figs. 5 and 6). However, the less-reworked nature of the Lake Johnston block at ∼2.7 Ga (Fig. 4) suggests that the ∼2.9 Ga Hyden block was thicker than this part of the ∼2.7 Ga West Yilgarn block (Figs. 5 and 6); an inference supported by magnetotelluric data (38). Consequently, at ∼2.9 Ga, the >150-km-thick lithosphere of the Hyden block prevented the segregation of high-percentage (∼50–30%) partial melt Munro-type magmas (MH) and restricted melt generation to Barberton- and low-percentage partial melt (<30%) Munro-type magmas (ML; Fig. 5). The lack of large quantities of Munro-type melt, together with the close proximity of the craton margin (Fig. 5), ensured Barberton-type melts were erupted in the Forrestania greenstone belt (17) before significant dilution by the upper high-percentage partial melt Munro-type magmas could occur. The eastward position of the Lake Johnston greenstone belt allowed homogenization of the Barberton–Munro melts, resulting in magmas with intermediate compositions (19) (Fig. 5).
In contrast, at ∼2.7 Ga in the Eastern Goldfields, due to the shallower nature of the Lake Johnston block, Barberton-type magmas generated under the Hyden block would have had to travel ∼300–200 km laterally through both low- and high-percentage partial-melt Munro-type sources (Fig. 6). Subsequently, the Barberton-type signature was diluted to the extent that only Munro-type komatiites were erupted in the Eastern Goldfields (9).
A Phanerozoic analog for the relationship between lithospheric architecture and magmatism in the Yilgarn Craton at ∼2.9 and 2.7 Ga is the ∼17 Ma (31, 32) continental plume setting at Yellowstone in the western United States. The overlying lithosphere comprises the old, thick Archean Wyoming Craton to the east and the thinner Mesozoic–Paleozoic-accreted oceanic terranes to the west (29⇓–31). The westward movement of the North American plate (∼2 cm/y; ref. 31) traversed this lithospheric architecture over the stationary plume, resulting in variations in the character, frequency, and isotopic signature of volcanism (29, 31, 32). Under the younger, thinner accreted terranes in the west, the plume ascends to higher levels allowing the formation of large volumes of decompression melt (29, 31, 32, 34, 35) and a juvenile (εNd +4, εHf +10; ref. 29) volcanic sequence dominated by basaltic (including the Columbia River basalts; ref. 32) and minor felsic magmatism. This setting is a modern analog to the high-MgO, voluminous continent-edge–type komatiite volcanism observed in the Yilgarn Craton. To the east, where the Yellowstone plume impinges on thick, old Archean lithosphere, less decompression melting occurs. This relatively small amount of mantle melt infiltrates the overlying lithosphere, where it is contaminated by more unradiogenic continental material. This process leads to evolved (εNd −11, εHf −10; ref. 29) felsic-only volcanism, similar to that recorded in the Marda block at ∼2.7 Ga (24) (εHf −7 to −2; Table S2). This architecture is a modern analog to the intracontinent Archean settings, where komatiite magmatism is largely absent.
This study demonstrates that the dynamic evolution of the early continents controlled the location, geochemistry, metallogeny, and volcanology of komatiites. An analogous process continues to operate in the modern Earth (30, 32, 34), and has been fundamental to the transfer of deep mantle material to the continental crust, oceans, and atmosphere throughout the history of the planet.
Materials and Methods
We report here on new U-Pb geochronology (36 samples) and Lu-Hf isotopic data (84 samples) from the Yilgarn Craton of Western Australia (Tables S2 and S3). The U-Pb zircon geochronology was performed on the sensitive high-resolution ion microprobes at the John de Laeter Centre of Mass Spectrometry at Curtin University, Western Australia. Following precise dating of the magmatic and inherited zircon populations from sampled granites and felsic volcanics, >900 in situ Lu-Hf isotopic analyses were carried out at the Centre for the Geochemical Evolution and Metallogeny of Continents at Macquarie University in Sydney, Australia. These data were then processed and plotted as time-slice contour maps using ArcGIS (Figs. 2–4 and Fig. S5). For the complete methodology we refer the reader to the Supporting Information.
Acknowledgments
Adam Wilson and Alex Clarke-Hale are thanked for field support. This project was funded by Australian Research Council (ARC) Linkage Grants LP0776780 and LP100100647 with BHP Billiton Nickel West, Norilsk Nickel, St Barbara, and the Geological Survey of Western Australia (GSWA). The GSWA is acknowledged for sample provision and technical advice. C.L.K., S.S.R., and M.P.D. publish with permission of the Executive Director of the Geological Survey of Western Australia. S.J.B.’s contribution is supported by the Commonwealth Scientific and Industrial Research Organization (CSIRO) Minerals Down Under National Research Flagship. The Lu-Hf analytical data were obtained using instrumentation funded by Department of Education Science and Training (DEST) Systemic Infrastructure grants, ARC Linkage Infrastructure, Equipment and Facilities (LIEF), National Collaborative Research Infrastructure Strategy (NCRIS), industry partners, and Macquarie University. The U-Pb zircon geochronology was performed on the sensitive high-resolution ion microprobes at the John de Laeter Centre of Mass Spectrometry (Curtin University). This is contribution 456 from the ARC Centre of Excellence for Core to Crust Fluid Systems (CCFS) and 939 from the Centre for the Geochemical Evolution and Metallogeny of Continents (GEMOC) Key Centre.
Footnotes
↵1Present address: Earth Science and Resource Engineering, Commonwealth Scientific and Industrial Research Organization (CSIRO), Kensington, Perth, WA 6151, Australia.
- ↵2To whom correspondence should be addressed. E-mail: david.mole{at}csiro.au.
Author contributions: D.R.M., M.L.F., and J.M. designed the research project; D.R.M., N.T., S.S.R., and M.P.D. performed research; D.R.M., M.L.F., K.F.C., T.C.M., C.L.K., and S.J.B. analyzed data; D.R.M. wrote the paper; N.T., K.F.C., T.C.M., C.L.K., and J.M. performed regional geological analysis; S.S.R. and M.P.D. performed regional geological analysis and mapping; E.A.B. provided assistance with operation of analytical equipment and data reduction; and S.J.B. provided access to the CSIRO komatiite database.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1400273111/-/DCSupplemental.
References
- ↵
- ↵
- ↵
- Herzberg C,
- et al.
- ↵
- ↵
- Arndt NT,
- Barnes SJ,
- Lesher CM
- ↵
- ↵
- ↵
- ↵
- Barnes SJ
- ↵
- Barnes SJ,
- Fiorentini ML
- ↵
- ↵
- Shimizu K,
- Nakamura E,
- Maruyama S
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- Heggie GJ,
- Fiorentini ML,
- Barnes SJ,
- Barley ME
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- Hawkesworth CJ,
- et al.
- ↵
- ↵
- Champion DC,
- Cassidy KF
- ↵
- Mole DR,
- et al.
- ↵
- ↵
- ↵
- ↵
- Hanan BB,
- Shervais JW,
- Vetter SK
- ↵
- Barnes SJ
- ↵
- Begg GC,
- et al.
- ↵
- ↵
- Witt WK
- ↵
- Robin-Popieul CC,
- et al.
- ↵
- Dentith MC,
- et al.
- ↵
- ↵
- Cassidy KF,
- et al.
- ↵
- ↵
- ↵
- Begg GC,
- et al.
Citation Manager Formats
Article Classifications
- Physical Sciences
- Earth, Atmospheric, and Planetary Sciences