Convective isolation of reservoirs through time

Jonas Tuscha,1, Carsten Münkera, Eric Hasenstaba, Mike Jansena, Chris S. Mariena, Florian Kurzweila, Martin J. Van Kranendonkb,c, Hugh Smithiesd, Wolfgang Maiere, and Dieter Garbe-Schönbergf

aInstitut für Geologie und Mineralogie, Universität zu Köln, 50674 Köln, Germany; bSchool of Biological, and Environmental Sciences, The University of New South Wales, Kensington, NSW 2052, ; cAustralian Center for Astrobiology, The University of New South Wales, Kensington, NSW 2052, Australia; dDepartment of Mines, Industry Regulations and Safety, Geological Survey of , East Perth, WA 6004, Australia; eSchool of Earth and Sciences, Cardiff University, Cardiff CF10 3AT, United Kingdom; and fInstitut für Geowissenschaften, Universität zu Kiel, 24118 Kiel, Germany

Edited by Richard W. Carlson, Carnegie Institution for Science, Washington, DC, and approved November 18, 2020 (received for review June 19, 2020) Although Earth has a convecting mantle, ancient mantle reservoirs anomalies in rocks was interpreted as evidence that that formed within the first 100 Ma of Earth’s history (Hadean these rocks lacked a late veneer component (5). Conversely, the Eon) appear to have been preserved through geologic time. Evi- presence of some late accreted material is required to explain the dence for this is based on small anomalies of isotopes such as elevated abundances of highly siderophile elements (HSEs) in 182W, 142Nd, and 129Xe that are decay products of short-lived nu- Earth’s modern silicate mantle (9). Notably, some Archean rocks clide systems. Studies of such short-lived isotopes have typically with apparent pre-late veneer like 182W isotope excesses were focused on geological units with a limited age range and therefore shown to display HSE concentrations that are indistinguishable only provide snapshots of regional mantle heterogeneities. Here from modern mantle abundances (10, 11), which is difficult to 182 182 we present a dataset for short-lived Hf− W (half- 9 Ma) in reconcile with the missing late veneer hypothesis. An alternative a comprehensive suite from the , Western Aus- suggestion is therefore that early silicate differentiation during tralia. The samples analyzed preserve a unique geological archive 182 182 the lifetime of Hf might have caused the formation of mantle covering 800 Ma of Archean history. Pristine W signatures that reservoirs with anomalous 182W signatures (10, 12). In addition, directly reflect the W isotopic composition of parental sources are recent studies have revealed variable 182W isotope deficits in the only preserved in unaltered mafic samples with near canonical mantle plume sources of ocean island (13). In line with

W/Th (0.07 to 0.26). Early , mafic igneous rocks from EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES noble gas systematics and seismic properties of such deep-rooted the East Pilbara Terrane display a uniform pristine μ182W excess of mantle plumes, these 182W deficits have been taken as evidence 12.6 ± 1.4 ppm. From ca. 3.3Ga onward, the pristine 182W signa- tures progressively vanish and are only preserved in younger rocks for the presence of primordial domains that have been con- vectively isolated since Earth’s earliest history (13). The origin of of the craton that tap stabilized ancient . Given that the – anomalous 182W signature must have formed by ca. 4.5 Ga, the this modern igneous reservoir remains ambiguous (14 16), and mantle domain that was tapped by magmatism in the Pilbara Cra- its presence in pre- time is highly speculative. ton must have been convectively isolated for nearly 1.2 Ga. This However, the contribution of modern mantle plume sources with 182 finding puts lower bounds on timescale estimates for localized W deficits to the convecting mantle offers an additional ex- ENVIRONMENTAL SCIENCES 182 convective homogenization in ’s interior and on the planation for the decrease of W isotope excesses since the widespread emergence of that are both important Archean as a consequence of increasing mantle homogenization input parameters in many physical models. (SI Appendix).

late veneer | | tungsten isotopes | early Earth | mantle Significance convection Geological processes like mantle convection or plate tectonics mong the terrestrial planets, Earth is unique in that plate are an essential factor controlling Earth’s habitability. Our Atectonic processes efficiently mix and homogenize its silicate study provides insights into timescales of convective homo- mantle. Surprisingly, however, geochemical studies have revealed genization of Earth’s early mantle, employing the novel tool of that both Archean and Phanerozoic mantle reservoirs still carry high-precision 182W isotope measurements to rocks from the primordial geochemical signatures, thus escaping efficient con- Pilbara Craton in Australia, that span an age range from 3.5 182 vective homogenization as also predicted by geodynamic models billion to 2.7 billion years. Previous W studies mostly (1, 2). The main evidence for such ancient geochemical hetero- covered snapshots through geologic time, so the long-term 182 geneities stems from noble gas systematics (3) and from short- W of the mantle has been ambiguous. Together lived nuclide decay series that became extinct after the Hadean with sophisticated trace element approaches, we can now eon (>4.0 Ga) (4, 5). For instance, the relative abundance of provide an improved insight into such timescales, arguing for daughter isotopes from short-lived nuclide series such as 142Nd local preservation of primordial geochemical heterogeneities within Earth’s mantle as late as around 3.0 billion years, the and 182W shows significant variations in ancient rocks when putative onset of widespread plate tectonics on Earth. compared to Earth’s modern mantle composition (4, 5). From 182 –182 these short-lived isotope systems, the Hf W decay system Author contributions: J.T. and C.M. designed research; J.T. performed research; J.T., M.J., has proven particularly useful in constraining the timing of C.S.M., W.M., and D.G.-S. contributed new reagents/analytic tools; J.T. analyzed data; and formation (6), timescales of late , and J.T., C.M., E.H., F.K., M.J.V.K., and H.S. wrote the paper. silicate differentiation (7, 8). The authors declare no competing interest. 182 There are two competing explanations for the origin of W This article is a PNAS Direct Submission. isotope anomalies found in the terrestrial rock record, arising Published under the PNAS license. from the markedly different geochemical behavior of Hf and W 1To whom correspondence may be addressed. Email: [email protected]. during both core formation and silicate differentiation in plan- This article contains supporting information online at https://www.pnas.org/lookup/suppl/ 182 etary bodies. As primitive meteorites exhibit strong W isotope doi:10.1073/pnas.2012626118/-/DCSupplemental. 182 182 deficits (μ W = −190) (6), the observation of positive W Published December 21, 2020.

PNAS 2021 Vol. 118 No. 2 e2012626118 https://doi.org/10.1073/pnas.2012626118 | 1of6 Downloaded by guest on September 28, 2021 Surprisingly few studies have directly assessed the 182W record supersuites of sodic (tonalites, trondhjemites, and granodiorites) of mantle-derived rock assemblages that span a relatively long and potassic granitoids () that can be regarded as probes time frame of Archean geodynamic evolution. Arguably, there of early mafic . The younger evolution of the EPT involved are already comprehensive studies available from the Slave plume-initiated rifting (Soanesville Group) at 3.18 Ga and sub- Craton (17), southwest (12, 18), and southern sequent accretion (3.07 Ga) of the younger WPS that also in- (19) that investigated lithostratigraphic successions covering time cludes ∼3.1-Ga -related mafic lithologies (Whundo series of several hundred million years. However, these studies Group) (30, 31). After amalgamation, postorogenic, lithosphere- only covered crustal reservoirs (Tonalite-throndhjemite-grano- derived magmatism included mafic rocks from the Bookingarra rocks [TTGs] and diamictites) but not mafic rocks that Group (Opaline Well Intrusion, Louden Volcanics, and Mount allow direct tracing of mantle-derived magmatic pulses. Notably, Negri Volcanics) and crust-derived posttectonic granites (e.g., fluid mobility of W already may have caused significant distur- Split Rock Supersuite) (27). bance of primary 182W isotope systematics in hydrated mafic We employed two strategies in selecting our samples. First, we precursors of such previously investigated TTG suites (17, 20, analyzed mafic volcanic rocks that tapped the ambient as- 21). Hence, W/Th ratios of TTGs may not necessarily reflect the thenospheric mantle of the EPT (plume derived) and the WPS primary W/Th ratios of their mafic precursors. In fact, sub- (subduction related). Secondly, to understand the evolution of canonical W/Th ratios are often observed in TTGs (e.g., ref. 22), the lithosphere and to obtain average crustal compositions, we that may reflect 1) preferential loss of W during dehydration at analyzed mafic dikes, , and granitoids of different ages. amphibolite facies conditions prior to partial anataxis or 2) We studied a total of 56 samples from more than 10 major preferential retention of W in residual rutile (23). It is therefore stratigraphic units of the two different terranes (SI Appendix, Fig. more than plausible that apparently invariant 182W isotope ex- S1), also covering different tectonic regimes (vertical tectonics in cesses in crustal assemblages (17) may simply mirror fluid- the EPT and horizontal tectonics in the WPS). For all these mediated W redistribution within their hydrated mafic precur- samples, selected trace element abundances (Zr, Nb, Ta, W, Th, sors (20, 21), obscuring primary magmatic trends. Moreover, and U) were obtained by isotope dilution measurements (details 182W studies on sediments like glacial diamictites only provide in SI Appendix). Ten ultramafic () samples were also average crustal compositions and protracted information on previously analyzed for HSE (32). High-precision W isotope convective processes in the mantle. An exceptional case in this measurements were performed on a subset of 30 samples. We regard are studies from southwest Greenland (5, 12, 18). These followed a slightly modified analytical protocol of reference (18) 3.8- to 3.4-Ga-old mantle-derived rocks from southwest Green- as outlined in SI Appendix. Results of high-precision W isotope show uniform 182W excesses but decreasing 142Nd anoma- analyses are reported in the μ-notation (equivalent to parts per lies through time in the same lithostratigraphic successions (12, million) relative to National Institute of Standards and Tech- 18). This relationship suggests that an original decrease in 182W nology Standard Reference Material 3163 and given in Dataset was obscured by secondary W redistribution during metamor- S1 (session mean values) and Dataset S2 (single measurements). phism (18), as also suggested for Archean rocks from the Acasta The μ182W isotope compositions addressed in this manuscript Complex (24), from the North Craton (25), and always refer to the measured 182W/184W ratio that has been from the Superior Province (26). Collectively, these examples corrected for mass bias by using 186W/184W = 0.92767 (denoted illustrate that most studies have only provided snapshots of the “6/4”) (33). Major and trace element compositions and isotope 182W isotope evolution of the mantle beneath individual Ar- dilution data are reported in Dataset S3. chean (SI Appendix) and that robust information on the High-precision concentration measurements are required to 182W isotope evolution of the Archean mantle with time is scarce identify samples with primary W abundances, because W is and often ambiguous, due to secondary disturbance. In fact, only strongly fluid mobile when compared to elements of similar in- 15% of all available samples previously analyzed for 182W display compatibility (e.g., Th, U, and Ta) (34). Thus, W mobility during W abundances that are pristine with respect to an unmodified fluid alteration results in noncanonical W/Th ratios (34), whereas, mantle source (Fig. 3 and SI Appendix). In order to allow for a in undisturbed magmatic systems, the similarly high incompati- more comprehensive understanding of Archean geodynamic bility of W and Th results in canonical W/Th ratios between 0.09 evolution, studies are required to assess the 182W isotope evo- and 0.24 over a wide range of mafic to felsic melts (34). Our lution of a particular region over a long time period, with a focus Archean rock samples display primary magmatic trends for Th on mantle-derived rocks. versus Zr (Fig. 1A), but not for W versus Zr, as W was enriched in Here we present a nearly continuous 182W isotope record for many samples (Fig. 1B). Samples with canonical W/Th ratios, an Archean craton that spans an age range of 800 Ma, from the however, define a pristine magmatic trend (Fig. 1B), allowing us to Paleoarchean to the , and covers both mafic and restrict further consideration primarily to samples with 182Wiso- felsic compositions. We investigated the 182W isotope evolution tope composition unmodified by secondary processes. of the exceptionally well preserved 3.58- to 2.76-Ga-old Pilbara 182 Craton (Western Australia). Since much of this craton is only W Systematics of the Pilbara Craton slightly affected by metamorphism (27), many units within the The oldest plume-related mafic volcanic samples from the EPT lithostratigraphic successions are more likely to have preserved () all display resolvable 182W excesses. their primary W systematics than other more deformed cratons However, Warrawoona Group rocks, and a slightly older gab- of similar age. Therefore, our data may allow a much more ro- broic enclave from the Shaw Granitic Complex (Mount Webber bust assessment of 182W isotope evolution through the Archean. Gabbro; Pil17-07), with supracanonical W/Th ratios, display A detailed outline of the geologic evolution of the Pilbara Cra- markedly lower μ182W compositions when compared to samples ton is provided in SI Appendix. In short, mantle-derived (mafic with undisturbed elemental W budgets. Samples with primary W and ultramafic) magmatic sequences in the east (East Pilbara concentrations (canonical W/Th) display an average of μ182W = Terrane [EPT]; Warrawoona, Kelly, and Sulphur Springs +12.6 ± 1.4 ppm [95% CI, n = 8, also including previously groups) and west (West Pilbara Superterrane [WPS]; Ruth Well published data for two samples from the Apex (15); red Formation) of the craton cover an age range from 3.52 Ga to bar in Fig. 2], whereas rocks affected by secondary W enrichment 3.23 Ga and tapped rather depleted mantle sources that were display a resolvably lower excess of ∼+8.1 ± 1.4 ppm (95% CI, emplaced during distinct mantle plume events, similar to modern n = 8; gray bar in Fig. 2). Interestingly, the average 182W excess plateau basalts (28, 29). Burial of these mafic sequences, to- in the metasomatized samples is indistinguishable from the long- gether with older protocrust, caused melting to form four term lithospheric average (Fig. 2 and below). Samples from the

2of6 | PNAS Tusch et al. https://doi.org/10.1073/pnas.2012626118 Convective isolation of Hadean mantle reservoirs through Archean time Downloaded by guest on September 28, 2021 A B R²=0.02 Metasomatic W enrichment ] μg/g R²=0.73 Th [ W [μg/g]

crustal contamination


Zr [μg/g] Zr [μg/g] C This study 20 Mafic-ultramafic rocks (non-canonical W/Th) Mafic-ultramafic rocks (canonical W/Th) 15 Post-orogenic Bookingarra Group Warrawoona average

Shales 10 W (6/4) W

182 5 Previous studies μ Metasomatic signatures Archer et al., 2019 (unknown W/Th) 0 Pristine Rizo et al., 2019 (canonical W/Th) signatures -5 0.1 1 10 undisturbed samples W/Th (W/Th 0.04 - 0.30)

Fig. 1. Trace element variation diagrams showing Th and W vs. Zr contents in mafic−ultramafic rocks from the Pilbara Craton (A and B) and high-precision μ182W data for rocks from the Pilbara Craton versus W/Th ratios (C). All concentration measurements were conducted by isotope dilution. Also included are Pilbara samples that were previously analyzed for their 182W isotope composition (15, 36). (A) Nearly all Pilbara samples have preserved robust magmatic EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES differentiation trends for Th vs. Zr, indicating that Th was not significantly affected by secondary processes. Only the postorogenic samples from the Bookingarra Group were contaminated by crustal components, in line with previous studies (35). (B) In contrast to Th, the scattered concentrations of W mirror postmagmatic W enrichment by secondary processes. Primary magmatic differentiation trends are preserved in samples with canonical W/Th ratios. Notably, all samples from ref. 36 display strongly elevated W concentrations that indicate postmagmatic W enrichment. (C) Samples with undisturbed ele- mental W budgets (red squares, W/Th lower than 0.30; SI Appendix) display variable 182W isotope excesses that illustrate a decrease of 182W excesses with time (Fig. 2). Samples with elevated W/Th ratios (grey squares) were affected by secondary W enrichment through metasomatic agents that integrate the 182W isotope composition of the whole Pilbara Craton. Accordingly, these samples display intermediate 182W isotope excesses yielding the same average as

lithosphere derived samples (grey and blue bars, respectively). ENVIRONMENTAL SCIENCES

younger, plume-derived ∼3.35-Ga Kelly Group (Euro Basalt) ) display μ182W values that are indistinguishable from the exhibit a contrasting pattern. The altered sample Pil16-39a dis- Warrawoona Group average (only considering samples with ca- plays a μ182W value of +8.5 ± 2.1 ppm, again indistinguishable nonical W/Th ratios). However, highly fractionated postorogenic from the long-term lithospheric average. However, the unaltered monzogranites (Pil16-36A, 2.85-Ga Spearhill Monzogranite and sample Pil16-53a shows a resolvably lower excess of μ182W = Pil16-12, 3.45-Ga North Pole Monzogranite; blue squares in +5.2 ± 2.6 ppm when compared to unaltered samples from the Fig. 2) display significantly lower μ182W values than this War- Warrawoona Group. The youngest sample from the EPT, a 3.18- rawoona Group average. Hence, the granitoids in the EPT mimic Ga Honeyeater basalt, displays no 182W isotope excess, over- the 182W isotopic evolution of the ambient mantle, although with lapping the modern upper mantle value (Pil16-51, μ182W = some delay. Given that the oldest and less evolved sodic granites +1.1 ± 3.9 ppm). The eruption of the Honeyeater Basalt marks are directly derived from mafic protocrust, this protocrust must extension at the onset of clear plate tectonic style processes (the have had an average 182W excess similar to that of the Warra- earliest recorded ) in the Pilbara Craton (31). Thus, woona Group samples. Lower 182W isotope excesses in younger decreasing μ182W values in the unaltered samples indicate a and more evolved granitoids mirror the temporal decrease in diminishing 182W isotope excess with time during a change in the 182W as also observed for mantle-derived mafic rocks. Prevalent tectonic regime. Additional evidence for change in geodynamic 182W excesses that are lower than the pristine Warrawoona pattern is manifested in rocks from the 3.13- to 3.11-Ga Whundo Group average (canonical (W/Th) are also recorded in Meso- Group (WPS), a sequence with geochemical characteristics that archean to Neoarchean shales (Pil16-50B, 3.19-Ga Paddy Mar- are consistent with derivation from a depleted mantle source that ket Formation and Pil16-38b, 2.76-Ga Hardey Formation, μ182W was variably affected by subduction components in an arc-like between +7.1 and +7.6 ppm), a -related lithosphere-derived setting (30). Samples from the Whundo Group exhibit consis- dolerite from the Fortescue Group (μ182W = +7.6 ± 3.1 ppm tently lower μ182W (Pil16-67: μ182W = +5.3 ± 4.8; Pil16-74: Pil16-31, 2.78-Ga Black Range dolerite) and mafic rocks from μ182W = +5.9 ± 3.2) than Warrawoona Group samples with the Bookingarra Group (ca. 2.95 Ga; Pil16-63, Pil16-65, and canonical W/Th ratios. Beside data for 3.48-Ga from Pil16-75), which are derived from a metasomatized, lithospheric the Komati Formation of the Kapvaal Craton (10), our data mantle source that was refertilized by subducted sediments in a provide the oldest evidence for modern upper mantle-like W subduction zone-like setting (35) (μ182W between +8 and +11 isotope compositions in the early Archean rock record. ppm; Fig. 2). The shales provide an average of the 182Wisotope The 182W isotope excesses also slightly decrease in granitoids: composition of the upper crust, and the lithosphere-derived mafic Sodic and potassic granitoids (Pil16-34 and Pil16-35, 3.47-Ga intrusion probes the composition of the subcontinental litho- North Shaw Tonalite and Pil16-41, 3.47-Ga Homeward Bound spheric keel. Unlike the shales, a Paleoarchean Algoma-type

Tusch et al. PNAS | 3of6 Convective isolation of Hadean mantle reservoirs through Archean time https://doi.org/10.1073/pnas.2012626118 Downloaded by guest on September 28, 2021 Terrane accretion altered samples were studied, we can now distinguish between 182 Ruth Well Whundo W - Pilbara diminishing W isotope anomalies in the ambient mantle and 182 Warrawoona KSSS E - Pilbara secondary processes that redistribute W isotope anomalies Mt Webber Tabletop Coucal Double Bar North Star Mt Ada Apex Euro Honeyeater (Fig. 1C). Warrawoona group (canonical W/Th) Our study demonstrates that the oldest mantle-derived rocks μ¹⁸²W = +12.6 ± 1.4 (95% CI) 182 R from the Pilbara Craton with canonical W/Th display W iso- 15 High μ¹⁸²W R tope excesses of a magnitude similar to other Archean cratons. As outlined above, however, there is no uniformly accepted ex-

10 planation for the origin of these 182W excesses. For the Pilbara W (6/4) W ¹⁸² Craton, the missing late veneer hypothesis provides a plausible

182 Metasomatic μ W

μ 5 explanation, as decreasing depletions of platinum group ele- ments (PGE) in komatiites with decreasing age are consistent Low μ¹⁸²W 0 with a progressive in-mixing of late veneer material (32). Among Metasomatized samples (supracanonical W/Th) Lithospheric average the 10 komatiite samples from the EPT and WPS that were μ¹⁸²W = +8.1 ± 1.4 (95% CI) μ¹⁸²W = +8.3 ± 1.0 (95% CI) -5 previously analyzed for HSE (32), we found only one sample 3.58 3.52 3.50 3.48 3.463.35 3.20 3.00 2.80 (179738) that preserved its primary W budget. All other komati- Time [Ga] ites from this sample selection were metasomatically overprinted and exhibit strongly elevated W/Th ratios up to 51 (sample Mantle-derived rocks Lithosphere-derived rocks 160245). This confirms previous evidence that ultramafic rocks BIF Canonical W/Th Shales Black Range dyke are extremely susceptible to (fluid mediated) second- en- Supracanonical W/Th Bookingarra Group Granitoids richment of W(17), which causes decoupling of HSE and 182W Fig. 2. High-precision μ182W analyses for rocks from the Pilbara Craton systematics. The komatiite sample 179738 (3.46-Ga Apex Basalt presented in stratigraphic order. Associated uncertainties refer to the cor- of the Warrawoona Group) with a canonical W/Th ratio is responding 95% CIs of multiple digestions. Mantle-derived samples with strongly depleted in PGE (37) and displays an elevated 182Wof undisturbed elemental W budgets (having canonical W/Th ratios) are rep- +11.9 ± 2.9 ppm, consistent with a model assuming an incom- resented by red squares; mantle-derived samples that were affected by plete late veneer contribution to the mantle source of this sam- secondary W enrichment (supracanonical W/Th ratios) are shown as grey ple. Such an interpretation is also in line with the absence squares. Undisturbed Warrawoona Group samples display an average of 142 μ182W = +12.6 ± 1.4 ppm (95% CI, n = 8; red bar), whereas rocks that were of Nd anomalies in Pilbara rock samples that show significant affected by secondary W enrichment display resolvably lower excesses of PGE depletions (36), thereby making unlikely early silicate ∼+8.1 ± 1.4 ppm (95% CI, n = 8; gray bar). Younger mantle-derived rocks crystal−liquid differentiation as an additional explanation for the 182 with undisturbed W budgets display significantly lower 182W excesses. occurrence of W isotope excesses in the Pilbara samples. Lithosphere-derived rocks are shown as blue symbols. Granitoid rocks (blue Collectively, our data indicate that Hadean, pre-late veneer squares) include nongneissic tonalites and more fractionated and post- mantle was preserved in the mantle sources of Pilbara igneous orogenic granitoids. Shales (blue circles) provide information about the rocks until ca. 3.3 Ga to 3.1 Ga, much longer than evident from upper , whereas lithospheric mafic magmatism is recorded other cratons (12, 26). The long-standing apparent mismatch in samples from the Bookingarra and Fortescue groups. The lithospheric 182 182 between virtually constant W isotope excesses and progres- samples provide an average μ W of +8.3 ± 1.0 ppm (blue bar) which is 142 significantly lower than the average μ182W of +12.6 ± 1.4 ppm for Warra- sively vanishing Nd anomalies through Archean time (12) is woona Group samples with canonical W/Th (red bar). Notably, samples that therefore well explained by the larger mobility of W compared to were affected by secondary W enrichment (gray bar) display the same av- Nd, and its redistribution during secondary processes. By only erage as the lithosphere-derived rocks (blue bar). R, previously published considering samples with strictly canonical W/Th ratios, we show data for two samples from the Apex basalt (15); K, Kelly Group; SS, Sulphur that the prevailing 182W isotope excesses even in the younger Springs Group; S, Soanesville Group. Archean rock record (Fig. 3) may be a vestige of older rocks, from which W was progressively redistributed. Hence, we rather argue that mantle convection homogenized primordial 182W banded formation from the EPT (Pil16-62B, 3.47-Ga Duffer isotope anomalies during the Paleoarchean and 142 Formation) has negligible detrital components (0.4% Al2O3), and (Fig. 3) at the same timescales as Nd systematics (38). 182 ± carries a W excess of +9.0 3.3 ppm. We interpret this sig- Geodynamic Implications as being seawater-derived and reflecting the average 182 composition of the ambient hydrothermal input provided by local Our W results for the Pilbara Craton have important impli- volcanism with similar W isotope composition. cations for understanding timescales of geodynamic processes on Altogether, the integrated 182W isotope compositions recor- early Earth and provide a future reference parameter for geo- ded in granitoids, shales, and lithosphere-derived rocks provide a dynamic models addressing mantle mixing in early Earth. Geo- lithospheric average μ182W of +8.3 ± 1.0 ppm (n = 11; blue bar dynamic models addressing the Pilbara data have to account for 182 the observations that 1) a Hadean pre-late veneer signature re- in Fig. 2), which is significantly lower than the average μ Wof mains isolated for more than 1 billion years and 2) such an +12.6 ± 1.4 ppm for Warrawoona Group samples with canonical isotopically isolated mantle domain continuously triggered pro- W/Th (n = 8; red bar in Fig. 2) but higher than the modern upper μ182 ± longed plume-driven magmatism with a distinct isotopic signa- mantle-like W of +1.1 3.9 ppm for the Honeyeater Basalt. ture, most likely in a stagnant lid regime (39). In contrast to Thus, the lithospheric average seems to record the integrated 182 182 present-day mantle plumes with negative W signatures (13), W isotope composition of different mantle reservoirs. Sam- Archean mantle plumes beneath the Pilbara Craton appear to ples from the EPT that were affected by secondary W enrichment have carried rather uniform 182W isotope excesses that were (grey bar in Fig. 2) show an average similar to the lithosphere- similar to the shallower mantle, thus arguing that composition- derived rocks, which is evidence for a craton-wide homogenization ally distinct deep mantle sources such as large low-shear velocity 182 of ambient W isotope compositions and the partial homo- provinces or ultra-low-velocity zones may not have been involved genization of primary 182W isotope variability. Hence, using or, alternatively, even may not have formed (39, 40). elemental W-Th systematics is a key tool, as this can unambiguously The marked geodynamic transition at 3.3 Ga to 3.1 Ga in the clarify the origin of 182W isotope variability. In contrast to a Pilbara Craton with influx of modern-like, upper mantle material previous study (36), where this tool has not been applied and carrying smaller 182W isotope excesses coincides with models

4of6 | PNAS Tusch et al. https://doi.org/10.1073/pnas.2012626118 Convective isolation of Hadean mantle reservoirs through Archean time Downloaded by guest on September 28, 2021 the mantle but were rather prolonged (38). The Archean re- gime was possibly characterized by small, isolated convection

Acasta LabradorNuvvuagittuqIsua SchapenburgPilbaraBarbertonIsua KostomukshaBoston Creek Phanerozoic cells and episodic mantle overturn events that vertically ho- samples mogenized discrete regions of the mantle (39, 45) but main- Archean Phaneroz. tained lateral heterogeneities for billions of years (2). Most 20 Pre-3.4 Ga Archean average likely, the initiation of modern plate tectonics was not a sudden μ¹⁸²W = +12.7 ± 1.2 (95% CI) event but rather occurred gradually (46). These circumstances might explain why 1) other Archean lithostratigraphic succes- 10 sions of similar age to the Pilbara Craton display invariant 182W patterns (17, 24) or already show modern upper mantle-like 0 μ182 182

W (6/4) W W isotope compositions (10) and, 2) in some cases, W

182 anomalies survived at least until the Late Archean (47). In fact, μ -10 our interpretation is in line with a recent study interpreting heterogeneous 182W isotope compositions in Archean rocks as

-20 expressing localized subduction events (as in the Pilbara Cra- ton) during a transition period (∼3.6 Ga to 2.7 Ga) from plume 4000 3500 3000 200 100 0 tectonics to modern plate tectonics (40). Time [Ma] In summary, our study demonstrates that the Pilbara Craton preserves a natural laboratory for studying Archean geodynamics Fig. 3. Compilation illustrating the secular 182W isotope evolution of the that may serve as a future test case for interdisciplinary studies 182 terrestrial mantle. This dataset includes all available W isotope literature assessing timescales of mantle dynamics through deep time. data for mantle-derived rocks. For Nuvvuagittuq, we assume a minimum Complementary to attempts trying to infer timescale estimates emplacement age of 3.75 Ga (50), being well aware that it might be older (51). The literature data are subdivided into samples with overprinted ele- for mantle stirring times from global modeling (48), our ap- mental W budgets (grey) and samples with canonical W/Th (red). Symbols proach adds to a more detailed information pattern providing with no color fill refer to samples with unknown W/Th ratios. Data for Pil- individual homogenization histories of isotopically anomalous bara rocks presented in this study are squares with thick black frames. Ref- Archean mantle domains. Therefore, the rock record of the erences and information on the data compilation are provided in Dataset S4. Pilbara Craton provides important observational constraints on Error bars are omitted for visual clarity, but uncertainties are given in convective processes within the Archean mantle which are in- EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES Dataset S4. dispensable for numerical simulations, as such input parameters define the evolutionary pathway of computational convection claiming a global transition from stagnant lid-type regimes to- models (49). ward modern plate tectonics. This transition is, for instance, postulated from crustal growth curves based on the record Data Availability. All study data are included in the article and (41), from the increased occurrence of subduction-like magma- supporting information. tism in the Archean (42), from the change of composition in the subcontinental lithospheric mantle (43), and from increasingly ACKNOWLEDGMENTS. This research was funded by the German Research ENVIRONMENTAL SCIENCES Foundation (Grant MU 1406/18-1) to C.M. as part of the Priority Program larger volumes of exposed felsic crust (44). It is therefore tempting 1833, “Building a Habitable Earth.” M.J.V.K. is supported by the Australian to postulate that the initiation of plate tectonics caused a more Research Council through Discovery Project DP180103204. H.S. publishes efficient homogenization of Earth’s mantle than vertical, with the permission of the Executive Director, Geological Survey of Western plume-driven dynamics that prevailed in a stagnant lid regime Australia. We thank Frank Wombacher for maintenance of the multicollec- as proposed for the older units of the Pilbara Craton. Indeed, tor-inductively coupled plasma-mass spectrometer and for managing the clean laboratory. We are grateful to Alessandro Bragagni, Frank Wombacher, geodynamic models for a stagnant lid regime show that, con- and Mario Fischer-Gödde for helpful discussions concerning analytical issues. trary to expectation, the timescales of mantle mixing in the Comments by two anonymous reviewers and the editor helped to improve the Archean were not accelerated due to a hotter thermal state of manuscript.

1. S. Labrosse, J. W. Hernlund, N. Coltice, A crystallizing dense magma ocean at the base 13. A. Mundl et al., Tungsten-182 heterogeneity in modern ocean island basalts. Science of the Earth’s mantle. Nature 450, 866–869 (2007). 356,66–69 (2017). 2. M. D. Ballmer, C. Houser, J. W. Hernlund, R. M. Wentzcovitch, K. Hirose, Persistence of 14. A. Mundl-Petermeier et al., Anomalous 182W in high 3He/4He ocean island basalts: strong silica-enriched domains in the Earth’s lower mantle. Nat. Geosci. 10, 236–240 (2017). Fingerprints of Earth’s core? Geochim. Cosmochim. Acta 271, 194–211 (2020). 3. H. Craig, J. E. Lupton, Primordial neon, helium, and hydrogen in oceanic basalts. Earth 15. H. Rizo et al., 182W evidence for core-mantle interaction in the source of mantle Planet. Sci. Lett. 31, 369–385 (1976). plumes. Geochem. Perspect. Lett. 11,6–11 (2019). 4. V. C. Bennett, A. D. Brandon, A. P. Nutman, Coupled 142Nd-143Nd isotopic evidence for 16. T. Yoshino, Y. Makino, T. Suzuki, T. Hirata, Grain boundary diffusion of W in lower Hadean mantle dynamics. Science 318, 1907–1910 (2007). mantle phase with implications for isotopic heterogeneity in oceanic island basalts by 5. M. Willbold, T. Elliott, S. Moorbath, The tungsten isotopic composition of the Earth’s core-mantle interactions. Earth Planet. Sci. Lett. 530,1–9 (2019). mantle before the terminal bombardment. Nature 477, 195–198 (2011). 17. J. R. Reimink et al., Tungsten isotope composition of Archean crustal reservoirs and 6. T. Kleine, C. Münker, K. Mezger, H. Palme, Rapid accretion and early core formation implications for terrestrial μ182W evolution. Geochem. Geophys. Geosyst. 21,1–16 on asteroids and the terrestrial planets from Hf-W chronometry. Nature 418, 952–955 (2020). (2002). 18. J. Tusch et al., Uniform 182W isotope compositions in Eoarchean rocks from the Isua 7. T. S. Kruijer, T. Kleine, M. Fischer-Gödde, P. Sprung, Lunar tungsten isotopic evidence region, SW Greenland: The role of early silicate differentiation and missing late ve- for the late veneer. Nature 520, 534–537 (2015). neer. Geochim. Cosmochim. Acta 257, 284–310 (2019). 8. M. Touboul, I. S. Puchtel, R. J. Walker, Tungsten isotopic evidence for disproportional 19. A. Mundl, R. J. Walker, J. R. Reimink, R. L. Rudnick, R. M. Gaschnig, Tungsten-182 in late accretion to the Earth and . Nature 520, 530–533 (2015). the upper continental crust: Evidence from glacial diamictites. Chem. Geol. 494, 9. C.-L. Chou, Fractionation of siderophile elements in the earth’s upper mantle. Proc. 144–152 (2018). Lunar Planet. Sci. Conf. 9, 219–230 (1978). 20. J. R. Reimink, T. Chacko, R. A. Stern, L. M. Heaman, The birth of a cratonic nucleus: 10. M. Touboul, I. S. Puchtel, R. J. Walker, 182W evidence for long-term preservation of Lithogeochemical evolution of the 4.02-2.94 Ga Complex. early mantle differentiation products. Science 335, 1065–1069 (2012). Res. 281, 453–472 (2016). 11. J. van de Löcht et al., Earth´s oldest mantle peridotites show entire record of late 21. J. R. Reimink et al., No evidence for Hadean continental crust within Earth’s oldest accretion. Geolog. Soc. Amer. 46, 199–202, https://doi.org/10.1130/G39709.1 (2018). evolved rock unit. Nat. Geosci. 9, 777–780 (2016). 12. H. Rizo et al., Early Earth differentiation investigated through 142Nd, 182W, and highly 22. M. R. Mohan, B. S. Kamber, S. J. Piercey, Boron and arsenic in highly evolved Archean siderophile element abundances in samples from Isua, Greenland. Geochim. Cosmo- felsic rocks: Implications for Archean subduction processes. Earth Planet. Sci. Lett. 274, chim. Acta 175, 319–336 (2016). 479–488 (2008).

Tusch et al. PNAS | 5of6 Convective isolation of Hadean mantle reservoirs through Archean time https://doi.org/10.1073/pnas.2012626118 Downloaded by guest on September 28, 2021 23. S. Klemme, S. Prowatke, K. Hametner, D. Günther, Partitioning of trace elements 36. G. J. Archer et al., Lack of late-accreted material as the origin of 182W excesses in the between rutile and silicate melts: Implications for subduction zones. Geochim. Cos- Archean mantle: Evidence from the Pilbara craton, western Australia. Earth Planet. mochim. Acta 69, 2361–2371 (2005). Sci. Lett. 528, 115841 (2019). 24. J. R. Reimink et al., Petrogenesis and tectonics of the Acasta Gneiss complex derived 37. W. D. Maier et al., Progressive mixing of meteoritic veneer into the early deep 142 182 from integrated petrology and Nd and W extinct nuclide-geochemistry. Earth mantle. Nature 460, 620–623 (2009). – Planet. Sci. Lett. 494,12 22 (2018). 38. V. Debaille et al., Stagnant-lid tectonics in early Earth revealed by 142Nd variations in 25. J. Liu, M. Touboul, A. Ishikawa, R. J. Walker, D. Graham Pearson, Widespread tung- late Archean rocks. Earth Planet. Sci. Lett. 373,83–92 (2013). sten isotope anomalies and W mobility in crustal and mantle rocks of the Eoarchean 39. J. H. Bédard, Stagnant lids and mantle overturns: Implications for Archaean tectonics, Saglek Block, northern Labrador, : Implications for early Earth processes and magmagenesis, crustal growth, mantle evolution, and the start of plate tectonics. W recycling. Earth Planet. Sci. Lett. 448,13–23 (2016). Geosci. Front. 9,19–49 (2018). 26. M. Touboul, J. Liu, J. O’Neil, I. S. Puchtel, R. J. Walker, New insights into the Hadean 40. Q.-F. Mei, J.-H. Yang, Y.-F. Wang, H. Wang, P. Peng, Tungsten isotopic constraints on mantle revealed by 182W and highly siderophile element abundances of supracrustal rocks from the Nuvvuagittuq , Quebec, Canada. Chem. Geol. 383, homogenization of the Archean silicate Earth: Implications for the transition of tec- – 63–75 (2014). tonic regimes. Geochim. Cosmochim. Acta 278,51 64 (2019). 27. M. J. Van Kranendonk, R. Hugh Smithies, A. H. Hickman, D. C. Champion, Review: 41. E. A. Belousova et al., The growth of the continental crust: Constraints from zircon Hf- Secular tectonic evolution of Archean continental crust: Interplay between horizontal isotope data. Lithos 119, 457–466 (2010). and vertical processes in the formation of the Pilbara Craton, Australia. Terra Nov. 19, 42. J. A. Pearce, Geochemical fingerprinting of oceanic basalts with applications to 1–38 (2007). classification and the search for Archean . Lithos 100,14–48 28. R. H. Smithies, M. J. Van Kranendonk, D. C. Champion, It started with a plume–Early (2008). Archaean basaltic proto-continental crust. Earth Planet. Sci. Lett. 238, 284–297 (2005). 43. S. B. Shirey, S. H. Richardson, Start of the Wilson cycle at 3 Ga shown by diamonds 29. E. Hasenstab et al., Evolution of the early to late Archean mantle from Hf-Nd-Ce from subcontinental mantle. Science 333, 434–436 (2011). isotope systematics in basalts and komatiites from the Pilbara Craton. Earth Planet. 44. B. Dhuime, A. Wuestefeld, C. J. Hawkesworth, Emergence of modern continental Sci. Lett. 553, 116627 (2020). crust about 3 billion years ago. Nat. Geosci. 8, 552–555 (2015). 30. R. H. Smithies, D. C. Champion, M. J. Van Kranendonk, H. M. Howard, A. H. Hickman, 45. C. O’Neill, A. Lenardic, L. Moresi, T. H. Torsvik, C. T. A. Lee, Episodic Precambrian Modern-style subduction processes in the Mesoarchaean: Geochemical evidence from subduction. Earth Planet. Sci. Lett. 262, 552–562 (2007). – the 3.12 Ga Whundo intra-oceanic arc. Earth Planet. Sci. Lett. 231, 221 237 (2005). 46. T. Gerya, Geodynamics of the early Earth: Quest for the missing paradigm. Geology 31. M. J. Van Kranendonk, R. Hugh Smithies, A. H. Hickman, M. T. D. Wingate, S. Bodorkos, 47, 1006–1007 (2019). Evidence for Mesoarchean (∼3.2 Ga) rifting of the Pilbara craton: The missing link in an 47. I. S. Puchtel, J. Blichert-Toft, M. Touboul, R. J. Walker, 182W and HSE constraints from early Precambrian Wilson cycle. Precambrian Res. 177, 145–161 (2010). 2.7 Ga komatiites on the heterogeneous nature of the Archean mantle. Geochim. 32. W. D. Maier et al., Progressive mixing of meteoritic veneer into the early Earth’s deep Cosmochim. Acta 228,1–26 (2018). mantle. Nature 460, 620–623 (2009). 48. C. O’Neill, V. Debaille, The evolution of Hadean-Eoarchaean geodynamics. Earth 33. J. Völkening, M. Köppe, K. G. Heumann, Tungsten isotope ratio determinations by – negative thermal ionization mass spectrometry. Int. J. Mass Spectrom. Ion Process. Planet. Sci. Lett. 406,49 58 (2014). 107, 361–368 (1991). 49. M. B. Weller, A. Lenardic, Hysteresis in mantle convection: Plate tectonics systems. 34. S. König et al., The Earth’s tungsten budget during mantle melting and crust for- Geophys. Res. Lett. 39,3–7 (2012). mation. Geochim. Cosmochim. Acta 75, 2119–2136 (2011). 50. N. L. Cates, S. J. Mojzsis, Pre-3750 Ma supracrustal rocks from the Nuvvuagittuq su- 35. R. H. Smithies, D. C. Champion, S. S. , Evidence for early LREE-enriched mantle pracrustal belt, northern Québec. Earth Planet. Sci. Lett. 255,9–21 (2007). source regions: Diverse magmas from the c. 3·0 Ga Mallina Basin, Pilbara Craton, NW 51. J. O’Neil, R. W. Carlson, D. Francis, R. K. Stevenson, Neodymium-142 evidence for Australia. J. Petrol. 45, 1515–1537 (2004). Hadean mafic crust. Science 321, 1828–1831 (2008).

6of6 | PNAS Tusch et al. https://doi.org/10.1073/pnas.2012626118 Convective isolation of Hadean mantle reservoirs through Archean time Downloaded by guest on September 28, 2021