The chemistry of fine-grained terrigenous sediments ...The chemistry of ﬁne-grained terrigenous...

Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

ScienceDirect

Geochimica et Cosmochimica Acta 255 (2019) 247–264

The chemistry of fine-grained terrigenous sediments revealsa chemically evolved Paleoarchean emerged crust

Nicolas D. Greber a,⇑, Nicolas Dauphas b

aDepartement des Sciences de la Terre, Universite de Geneve, 1205 Geneve, SwitzerlandbOrigins Laboratory, Department of the Geophysical Sciences and Enrico Fermi Institute, The University of Chicago, Chicago, IL 60615, USA

Received 28 November 2018; accepted in revised form 9 April 2019; available online 18 April 2019

Abstract

The nature of the rocks exposed to weathering and erosion on continents exerts an important control on weathering feed-backs and the supply of nutrients to the oceans. It also reflects the prevailing tectonic regime responsible for the formation ofcontinents. How the chemical and lithological compositions of the continents evolved through time is, however, still a matterof debate. We use an extensive compilation of terrigenous sediment compositions to better constrain the nature of rocks at thesurface of continents at 3.25 Gyr and 250 Myr ago. Specifically, we use geochemical ratios that are sensitive indicators ofkomatiite, mafic, and felsic rocks in the provenance of the sediments. Our results show that the average Al2O3/TiO2 ratioof fine-grained terrigenous sediments decreased slightly over time from 26.2 ± 1.3 in the Archean to 22.1 ± 1.1 (2SE) inthe Phanerozoic. In contrast, in the same time interval, the average Zr/TiO2 ratio stayed nearly constant at �245. Consideringthe distinct behaviors of Al, Ti and Zr during sedimentary processes, we find that hydrodynamic mineral sorting had a minoreffect on the chemical composition of Archean fine-grained sediments, but could have been more effective during periods ofsupercontinents. We show that the compositions of Phanerozoic sediments (Al2O3/TiO2, Zr/TiO2, La/Sc, Th/Sc, Ni/Co, Cr/Sc) are best explained with igneous rocks at the surface of continents consisting of 76 ± 8 wt% felsic, 14 ± 6 wt% Arc-basaltsand 10 ± 2 wt% within-plate basalts, most likely in the form of continental flood basalts. Applying the same mass-balancecalculations to the Paleoarchean suggests continental landmasses with 65 ± 7 wt% felsic, 25 ± 6 wt% mafic and 11 ± 3 wt%ultramafic rocks (all 2SE), likely in the form of komatiites. The presence of volumetrically abundant felsic rocks at the surfaceof continents (as evident from the sediment record) as well as at mid-crustal levels (as evident from presently exposed igneousrock record) in Paleoarchean cratons is currently best explained with the onset of subduction magmatism before 3.25 Gyr.� 2019 Elsevier Ltd. All rights reserved.

Keywords: Sediment; Provenance; Zircon; Supercontinent; Continental Crust; Early Earth; Archean; Geodynamics

1. INTRODUCTION

In recent years, growing attention has been paid to theconnections between Earth’s mantle, the continental crust,and near surface environments. Fine-grained terrigenoussediments are key geological witnesses of how these reser-voirs evolved through time. For example, the chemical

https://doi.org/10.1016/j.gca.2019.04.012

0016-7037/� 2019 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: [email protected] (N.D. Greber).

and isotopic compositions of detrital sediments can be usedto quantify the redox state of the atmosphere and oceansthrough time (e.g. Canfield, 2005; Lyons et al., 2014, andreferences therein) as well as the composition of the conti-nental crust that was subjected to weathering and erosion(hereafter called ‘‘emerged crust”) (Taylor andMcLennan, 1985; Condie, 1993; Rudnick and Gao, 2003).Such studies have helped identify temporal and possiblycausal relationships between shifts in the composition andextent of the continental crust with changes in surface envi-ronments and habitats for life, such as the Great Oxidation


mailto:[email protected]


http://crossmark.crossref.org/dialog/?doi=10.1016/j.gca.2019.04.012&domain=pdf

248 N.D. Greber, N. Dauphas /Geochimica et Cosmochimica Acta 255 (2019) 247–264

Event (GOE) (Condie, 2005; Bindeman et al., 2016; 2018;Greber et al., 2017a; Smit and Mezger, 2017).

The goal of our study is to reconstruct the rock compo-sition of the emerged crust by using the chemical composi-tions of sediments. This task is challenging, as processessuch as chemical weathering, diagenesis and hydrodynamicsorting of minerals can bias the chemical and isotopic com-positions of sediments compared to those of their sourcerocks. In particular, elements that have high solubilities inwater like Mg, Ca, Sr, K and Na can be fractionated duringweathering and sedimentary processes (Taylor andMcLennan, 1985). Thus, one often relies on ratios of ele-ments that are mostly insoluble, and use these to estimatethe bulk composition of the emerged crust. Examples of ele-ment ratios that were used for this purpose are La/Sc,Th/Sc, Th/Co, Ni/Co, Cr/Zn, Th/Cr, La/Cr and Cr/U(Taylor and McLennan, 1985; Condie, 1993; Tang et al.,2016; Smit and Mezger, 2017; Large et al., 2018). Theobserved shifts in the element ratios involving either Nior Cr of fine-grained terrigenous sediments towards theArchean-Proterozoic boundary were used to argue for alargely mafic emerged crust prior to 3.0 Gyr, which wouldhave changed to be dominated by felsic rocks by around2.5 Gyr (Tang et al., 2016; Large et al., 2018). This changein the nature of the continental crust was used to argue forthe initiation of modern style plate tectonics at �3.0 Gyrago (Dhuime et al., 2015; Tang et al., 2016). It was also sug-gested that the fast transition from mafic to felsic rockscould have caused the first irreversible rise in atmosphericoxygen that occurred at around 2.5 Gyr ago (Lee et al.,2016; Smit and Mezger, 2017). However, this interpretationof the chemical composition of fine-grained terrigenous sed-iments was recently questioned based on the Ti isotopiccomposition of old sediments. As the Ti isotopic composi-tion of common igneous rocks correlates with the degreeof magmatic differentiation (Millet and Dauphas, 2014;Millet et al., 2016; Greber et al., 2017b; Deng et al.,2019), the Ti isotopic signature of fine-grained terrigenoussediments can be used as a proxy to reconstruct the chem-ical composition of the emerged crust (Greber et al., 2017a).In contrast to the proxies that rely on element ratios, nomajor change in the Ti isotopic composition of the sedi-ments was found since �3.5 Gyr ago, implying that theemerged crust contained >50 wt% felsic lithologies sincethen (Greber et al., 2017a). A potential explanation forthe inconsistencies between the different studies is that Niand Cr concentrations in sediments are not tracking theamount of exposed mafic crust, but record instead thegreater contribution of komatiites and ultramafic rocks inthe Archean compared to the Proterozoic and Phanerozoic(Greber et al., 2017a). Deng et al. (2019) argued insteadthat the heavy Ti isotopic composition of Archean terrige-nous sediments might be consistent with an emerged crustmade primarily of plume-related tholeiitic basalts and dif-ferentiated rocks akin to those found today in Hawaii orIceland. Further work is also needed to assess whether Tiisotope systematics can be affected by hydrodynamic grainsize sorting and weathering.

To evaluate the reasons for the observed discrepanciesbetween the different studies, we compiled Al, Ti and Zr

concentrations in fine-grained terrigenous sediments fromthe literature and use already published compilations ofCr, Ni, Co, Sc, Th and La concentrations of sediments(Taylor and McLennan, 1985; Tang et al., 2016; Greberet al., 2017a) to reconstruct the lithological compositionof the emerged continents 3.25 Gyr ago, a time of con-tention with regard to the nature of the emerged crustand for which a large enough database of fine-grained ter-rigenous sediments is available to draw meaningfulconclusions.

2. PROXY RATIOS AND MINERAL SORTING IN

SEDIMENTS

An element used to evaluate the provenance of fine-grained terrigenous sediments should ideally fulfill the fol-lowing criteria: (i) it should be fluid immobile and thusimmune to processes involving fluid-rock interaction, (ii)it should not be involved in biological processes, (iii) itsconcentration should be distinct among the differentlithologies in the provenance, and (iv) it should be mini-mally affected by mineral sorting during sedimentary pro-cesses. Aluminum, Ti and Zr are among the most fluidimmobile elements known (Taylor and McLennan, 1985)and are unimportant in biological processes. They thus ful-fill criteria (i) and (ii). Because Ti is overall more compatibleduring fractional crystallization than both Zr and Al, theAl2O3/TiO2 and the Zr/TiO2 ratios in igneous rocks corre-late positively with the SiO2 concentration, which is usedhere as proxy for the magmatic evolution of an igneous sys-tem (Fig. 1). Also, no major difference can be observed inthese geochemical trends between rocks of Archean andpost-Archean age as well as between samples belonging tothe alkaline, tholeiitic and calc-alkaline magmatic series(Fig. 1 and supplementary Fig. S1). Thus, the Al2O3/TiO2

and Zr/TiO2 ratios can be used to discriminate between fel-sic (63 < SiO2 < 80 wt%) and mafic components(45 < SiO2 < 52 wt%; MgO < 18 wt%). These elements thusfulfill criterion (iii). Regarding criterion (iv), Garcia et al.(1994) suggested that Al and Ti behaved similarly duringsediment transport and that their ratio should not beaffected by mineral sorting, allowing one to use theAl2O3/TiO2 ratio in fine-grained terrigenous sediments toevaluate the chemical and lithological composition of theirprovenance (Hayashi et al., 1997). Although publishedstudies indicate that these two elements do not get stronglydecoupled during sedimentary processes, Al is mainlyhosted in clay minerals, while Ti is also partly concentratedin resistant heavy minerals such as ilmenite or rutile. Thus,more work is needed to ascertain that the Al2O3/TiO2 ratiois not fractionated by mineral sorting. Zirconium can befractionated during sedimentary transport due to the pref-erential sorting of zircon grains into sandstones (Garciaet al., 1991; 1994; Garconet al., 2013b; 2014) and theZr/TiO2 ratio of fine-grained terrigenous sediments mightbe affected by mineral sorting during transport and sedimen-tation. The extent to which Al, Ti and Zr concentrations canbe fractionated by mineral sorting processes is currently dif-ficult to quantify and depends on several factors; mostimportantly the chemical and mineralogical composition

Fig. 1. Al2O3/TiO2 and Zr/TiO2 weight ratios vs. SiO2 concentrations of rocks from the dataset used in Keller and Schoene (2012), containinga large diversity of plutonic and volcanic lithologies. Only rocks with total oxides between 99 to 101 wt% are considered. A and B: Blue pointsrepresent samples younger than 2.5 Gyr and the orange points are samples older than 2.5 Gyr. C and D: Green points represent tholeiiticrocks and purple points are calc-alkaline samples (alkaline rocks are not shown). For more information about the definitions and filteringprocess for the different magmatic series see supplementary Fig. S1. For clarity, the y-axes for the Al2O3/TiO2 and Zr/TiO2 ratios were cut at120 and 1200, respectively. Both ratios correlate positively with SiO2 irrespective of their age and magmatic series. Thus, the Al2O3/TiO2 andZr/TiO2 ratios in fine-grained terrigenous sediments provide clues about the average rock composition of the sediment provenance. (Forinterpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

N.D. Greber, N. Dauphas /Geochimica et Cosmochimica Acta 255 (2019) 247–264 249

of the provenance, the fluvial transport distance, and thefluid motion pattern (Bouchez et al., 2011; Garconet al.,2013a). However, it is reasonable to assume that if any biasassociated with mineral sorting is present, Zr concentra-tions will be most affected, followed by Ti and then Al.As Al is enriched in clay minerals, the Al2O3/TiO2 ratioof fine-grained sediments should increase due to hydrody-namic mineral sorting, if it changes at all. In contrast, asZr is enriched in sandstones, the Zr/TiO2 and Zr/Al2O3

ratios of fine-grained sediments might decrease due to sed-imentary processes. This distinct behavior between theAl2O3/TiO2 and Zr/TiO2 ratios during sediment transportis corroborated by data from suspended and bedload sedi-ments from the Amazon and Ganga rivers (Bouchez et al.,2011; Lupker et al., 2011) (see supplementary Fig. S2), butthe question of to what extent such a bias impacts the geo-chemical record of fine-grained sediments on a 100 Myrtimescale needs to be addressed. Using the Al2O3, TiO2

and Zr concentrations together can thus provide cluesabout the chemical and lithological composition of a sedi-ment’s provenance as well as on the effect of mineral sortingon their chemical composition.

3. DATA COMPILATION, DATA TREATMENT, AND

RESULTS

Chemical compositions of fine-grained terrigenous sedi-ments were compiled from 63 publications (SupplementaryTable S1). This dataset comprises Al2O3 (wt%), TiO2 (wt%)and Zr (mg/g) concentrations of 1437 samples from 191 dif-ferent localities spanning ages from present to 3.8 Gyr. Q-Qplots show that the Al2O3/TiO2 and Zr/TiO2 weight ratiosdepart from normal distributions and are better explainedby log-normal distributions (supplementary Figs. S3 andS4). The Al2O3/TiO2 and Zr/TiO2 ratios were first filteredfor outliers using the Chauvenet criterion based on a log–normal distribution. Hereafter, this filtered dataset isnamed ‘‘individual shale dataset”. After this first outlierrejection step, the Al2O3/TiO2 and Zr/TiO2 ratios of sam-ples from the same locality and age were averaged to avoidoverrepresentation of localities with many samples overthose with only few samples. To remove Al2O3/TiO2 andZr/TiO2 ratios of sediments with an unusual provenanceand that do not represent a large-scale sampling of the con-tinents, the averaged ratios were also filtered for outliers


using the Chauvenet criterion based on a log–normal distri-bution. This dataset is named ‘‘averaged shale dataset”.

The Al2O3/TiO2 weight ratios (both Al2O3 and TiO2 inwt%) of the ‘‘averaged shale dataset” range from 38.4 to12.0 and decrease slightly with time from 26.2 ± 1.3 (2SE;n = 56) in the Archean to 22.1 ± 1.1 (2SE; n = 47) in thePhanerozoic (Figs. 2 and 3). The Zr/TiO2 weight ratios(Zr in ppm and TiO2 in wt%) show more relative dispersionthan the Al2O3/TiO2 ratios and range from 73 to 585(Figs. 2 and 3), but exhibit within error constant arithmetic

Fig. 2. Histograms and Kernel density estimations of filtered and localitygrained terrigenous sediments sorted by age. Indicated in the panels aremean, and (iii) the 2SE value. While the average Al2O3/TiO2 ratio is sligerror over time.

averages between the Archean (243 ± 24; 2SE, n = 55) andPhanerozoic (244 ± 27; 2SE, n = 42).

4. DISCUSSION

4.1. Workflow

We first discuss the potential impact of hydrodynamicgrain size sorting on the chemical composition of the sedi-ments by evaluating the relative abundances of their Al2O3,

averaged Al2O3/TiO2 (left) and Zr/TiO2 (right) weight ratios of fine-(i) the number of locations that passed all filtering tests (n), (ii) thehtly decreasing, the average Zr/TiO2 ratio does not change within

Fig. 3. Filtered and locality averaged Al2O3/TiO2 (A), Zr/TiO2 (B), Cr/Sc (C) and Ni/Co (D) weight ratios of fine-grained terrigenoussediments plotted versus their age. Cr/Sc and Ni/Co ratios are from the compilation of Greber et al. (2017a) and are location-averaged as wasdone for Al2O3/TiO2 and Zr/TiO2 ratios. We only consider Ni/Co and Cr/Sc ratios that passed the filter in Greber et al. (2017a), removingsamples affected by weathering. Diamictites are highlighted in red while other fine-grained terrigenous sediments are in blue. Indicated are thecalculated regressions (black line) and the 95% confidence interval of the mean (grey band). The correlation between the Al2O3/TiO2 ratio andage t (in Myr) of a sediment can be fitted with a power law Al2O3=TiO2ð ÞShale¼ 20:8036þ1:00059t. The Zr/TiO2 ratio of fine-grainedterrigenous sediments is rather constant and can be fitted with a linear regression Zr=TiO2ð ÞShale¼ 266:614-0:0067 � t. The Cr/Sc and Ni/Coratios are fitted with the following formulas: Cr=Scð ÞShale¼ 5:5253þ 6:160 � 10�10 � t3 and Ni=Coð ÞShale¼ 2:1216þ 1:400 � 10�10 � t3.


TiO2 and Zr concentrations. We then show how the Al2O3,TiO2 and Zr concentrations of fine-grained terrigenous sed-iments (and their metamorphosed equivalents) are trans-lated into the igneous rock composition of theirprovenance. To do so, we follow the modelling approachoutlined in Greber et al. (2017a), where a set of mass-balance equations are used to solve the measured composi-tions of the fine-grained terrigenous sediments for the con-tributions of different igneous rock endmembers. TheAl2O3/TiO2 and Zr/TiO2 ratios are only diagnostic of felsicvs. mafic rocks, so to tease apart the contribution ofkomatiites, we consider ratios involving elements that arerelatively compatible in pyroxene and olivine, such as Ni,Co, Cr and Sc (Adam and Green, 2006). All these proxiesare insensitive to the presence of chemical sediments likecarbonates and banded iron formations on the emergedlands, but as discussed, can be applied to evaluate the rela-tive contribution of different igneous rock types to the sed-iment record.

To test the applicability of the Al2O3/TiO2 and Zr/TiO2

ratios of fine-grained terrigenous sediments to reconstructthe composition of their provenance, we first apply ourapproach to the mid-Phanerozoic crust 250 Myr ago, a time

when there are good constrains on the nature of emergedcontinents. We then calculate the rock composition of theemerged crust 3.25 Gyr ago.

4.2. Evaluating the degree of hydrodynamic mineral sorting

over time

Aluminium, Ti and Zr concentrations of sediments havepreviously been used to evaluate the impact of mineral sort-ing processes on the chemical composition of sediments(Garcia et al., 1991; 1994; McLennan et al., 1993;Bouchez et al., 2011; Garzanti et al., 2011). Mineral sortingcan affect the composition of a sediment by removing denseminerals such as zircon grains into the coarse-grained sedi-ment fraction. On the other hand, recycling of such coarsegrained sediments like sandstones can also produce a fine-grained sediment fraction that is enriched in elements nor-mally associated with dense minerals (McLennan et al.,1993). As diamictites are glacial sedimentary deposits, ithas been suggested that they are generally less influencedby mineral sorting processes (Gaschnig et al., 2016). Whilethe Al2O3/TiO2, Ni/Co and Cr/Sc ratios of all sedimenttypes in our database largely overlap at any given time dur-

Fig. 4. Top: Zr/Al2O3 weight ratios of diamictite (red) and non-diamictite samples (blue) from the averaged shale database. Theblack line is a moving median with step size of 25 Myr and windowwidth of 250 Myr through the non-diamictite samples (blue circles)to investigate the impact of mineral sorting on their Zr/Al2O3 ratio.The dotted red line is the time dependent average composition ofigneous rocks (see Fig. S5). Bottom: detrital zircon age record afterVoice et al., (2011). Periods of supercontinent formation areindicated on top of the figure and are based on Cawood andHawkesworth (2015). Periods in which sediments display lowermedian Zr/Al2O3 ratios compared to the mean of igneous rocksseem to be correlated with peaks in the detrital zircon record andtimes of supercontinents. (For interpretation of the references tocolour in this figure legend, the reader is referred to the web versionof this article.)


ing Earth history (Fig. 3), diamictites seem to have on aver-age a higher Zr/TiO2 ratio compared to non-diamictite sed-iments (hereafter summarized as shales) and thisdiscrepancy is larger in the Proterozoic and Phanerozoiccompared to the Archean (Fig. 3B). This could mean thatshales and possibly diamictites have been affected by grainsize sorting. One of the most sensitive proxies to trackhydrodynamic mineral sorting should be the Zr/Al2O3

(ppm/wt%) weight ratio, as Al2O3 is enriched in finegrained clay minerals and Zr is enriched in dense zircongrains. Another advantage of this system is, that maficand felsic igneous rocks do not exhibit a strong differencein their Zr/Al2O3 ratios (supplementary Fig. S5). Conse-quently, if fine-grained terrigenous sediments were signifi-cantly impacted by mineral sorting processes, one wouldexpect to observe a difference between the average Zr/Al2O3 ratio of the sediments and that of the igneous rockrecord. We thus calculated the Zr/Al2O3 ratio of ouralready filtered and locality averaged shale database bydividing the Zr/TiO2 with the Al2O3/ TiO2 ratio (supple-mentary Table S1). In the Archean, sediments define arather narrow range in their Zr/Al2O3 ratio; diamictites(11.6 ± 2.4; n = 4), shales (9.5 ± 0.9; 2SE, n = 48) andigneous rocks (8.6 ± 0.45; 2SE, n = 2342) overlap withinerrors (Fig. 4). In the post-Archean, the scatter of the Zr/Al2O3 ratio of the sediment and the igneous rock recordbecomes more pronounced, especially after around1000 Ma (Figs. 4 and S5). On average, shales younger than1000 Myr yield a Zr/Al2O3 ratio (11.5 ± 1.1, 2SE, n = 61)that is around 9.5% lower, but still within error identicalto that of the contemporary igneous rock record (12.7± 0.24, n = 13,096). Diamictites on the other hand definea higher Zr/Al2O3 ratio (18.7 ± 3.0, n = 15) than igneousrocks. There are various possible explanations for theexcess Zr in diamictites, including that it is an inherited sig-nal from a local source environment that contains igneousrocks with high Zr/Al2O3 ratios or that the glaciers thatproduced these diamictites abraded and recycled coarsegrained and zircon-rich sediments (McLennan et al.,1993). To summarize, hydrodynamic mineral sorting hada seemingly small effect on the chemical composition ofArchean fine-grained sediments, but it might have becomemore important over time. The fine-grained sediments thatwere deposited during the past 1000 Myr show more varia-tion in their Zr/Al2O3 ratios. Our dataset is dominated bynon-diamictite sediments that seem to be slightly depletedin Zr when compared to the igneous rock record, as isexpected from the scavenging of zircon grains into thesand-sized sediment fraction. A consequence is that the lin-ear equation applied to the age versus Zr/TiO2 correlation(Fig. 3B) of the fine-grained sediment record is useful toidentify long-term trends in the lithologic composition ofthe emerged crust, but likely provides a minimum estimatefor the proportion of felsic rocks in a sediment provenance.

To investigate potential changes in the mode of hydro-dynamic mineral sorting over Earth’s history, we applieda moving median with a step size of 25 Myr and a uniformkernel window width of 250 Myr to the Zr/Al2O3 ratio ofthe non-diamictite sediments. We find several time intervalsof around 200 to 400 Myr width that are characterized by

median Zr/Al2O3 ratios that are lower than the averagecomposition of the igneous rock record (Fig. 4). Interest-ingly, the periods defined by low Zr/Al2O3 ratios correlatewith age peaks in the detrital zircon record of Voiceet al., (2011). It has been shown that these age peaks alsobroadly correlate with periods of supercontinents(Worsley et al., 1984; Campbell and Allen, 2008;Hawkesworth et al., 2009; Roberts and Spencer, 2015)(Fig. 4). The temporal correlation between peak ages indetrital zircon grains, low Zr/Al2O3 ratios in the fine-grained sediment record, and larger landmasses is in agree-ment with studies that suggest that transport distance andmineral sorting efficiency are positively correlated (Powell,1998; Garconet al., 2013b). It is currently debated if thedetrital zircon age peaks represent episodic crustal growthdue to enhanced magmatic activity (Arndt and Davaille,2013; Condie et al., 2017), or if they are a preservation bias


phenomenon, meaning that rocks related to the collisionalmountain building stages of the Wilson-Cycle are bettershielded from erosion and drainage into the ocean(Hawkesworth et al., 2009; Cawood and Hawkesworth,2015). We interpret the coincidence of periods with lowZr/Al2O3 ratios and zircon age peaks to reflect a sedimen-tological regime that allowed for an increased productionof sandstones and efficient hydrodynamic mineral sorting.In a supercontinent, transport distances from source rocksto marine sediments would be longer and there would bemore opportunities for large and dense minerals to betrapped on the continents, for example in foreland basinsand associated lake systems. More work is needed to ascer-tain the observation that supercontinents are associatedwith a decrease in the Zr/Al2O3 ratio of shales but the pre-sent study shows that subtle variations in the geochemistryof terrigenous sediments through time may convey impor-tant clues on sediment transport on continents.

Fig. 5. Zirconium concentration vs. Al2O3/TiO2 ratio in differenttypes of Phanerozoic mafic rocks. Empty circles are average valuesfor various continental flood basalts (LIPs; i.e., CAMP, Emeishan,Deccan Traps and Parana) calculated from the GeoRoc databaseand Hooper (2000). The square symbols represent differentestimates for Arc-basalts: the light grey square is the mean ofcontinental Arc-basalts and the dark grey square is the mean ofoceanic Arc-basalt (both from Kelemen et al. 2014) and the emptysquare is the average of Arc-basalt (both continental and oceanic)after the PetDB database. The average composition of MORBs isfrom Gale et al. (2013), and the two intra-continental basaltexamples (Chaıne des Puys, France; Doufutun, central China) arebased on Hamelin et al. (2009) and Li et al. (2015), respectively.The data have been filtered following the criteria outlined in Greberet al. (2017a), i.e. only rocks with SiO2 concentrations between 45and 52 wt% and MgO < 18wt% and total major element concen-trations between 99 and 101 wt% were considered. Errors are 2SEand sometimes smaller than the symbol size. For data andreferences, see Table 1.

4.3. Reconstructing the lithological composition of the mid-

Phanerozoic emerged crust

To unravel the nature of the igneous rocks exposed toweathering 250 Myr ago, we first need to define the rockendmembers. Igneous rocks can be broadly divided intoultramafic (<45 wt% SiO2), mafic (45–52 wt% SiO2;MgO < 18 wt%), intermediate (52–63 wt% SiO2) and felsicrocks (>63 wt% SiO2) (Le Bas and Streckeisen, 1991). Outof these groups, ultramafic rocks are expected to be of sub-ordinate importance for the Phanerozoic continents. Fur-thermore, as already indicated by their name, thechemical characteristics of intermediate rocks are inbetween those of mafic and felsic rocks. Regardless of theirformation mechanism, the element composition of interme-diate rocks like andesites can effectively be modeled as amixture of mafic and felsic rocks. Thus, they are notregarded as an individual endmember in our model. There-fore, to reconstruct the lithological composition of the mid-Phanerozoic emerged continents, we limit our endmembersto mafic (subscript M) and felsic rocks (subscript F). Themass fractions of felsic (f F) and mafic (fM) componentsin the 250 Myr old fine-grained terrigenous sediments(and by extension of their provenance) can be calculatedusing their Al2O3/TiO2 and Zr/TiO2 ratios and the mass-balance equations:

Al2O3

TiO2

� �shale

¼fF TiO2½ �F Al2O3

TiO2

� �Fþð1-fFÞ TiO2½ �M Al2O3

TiO2

� �M

fF TiO2½ �Fþð1-fFÞ TiO2½ �M;

ð1Þ

Zr

TiO2

� �shale

¼fF TiO2½ �F Zr

TiO2

� �Fþð1-fFÞ TiO2½ �M Zr

TiO2

� �M

fF TiO2½ �Fþð1-fFÞ TiO2½ �M;

ð2ÞGreber et al. (2017a) used the Ti isotopic composition of

fine-grained terrigenous sediments to calculate theproportions of felsic and mafic rocks in the mid-Phanerozoic emerged crust using a similar mass-balanceequation:

d49Tishale ¼ fF TiO2½ �Fd49TiFþð1-fFÞ TiO2½ �Md49TiMfF TiO2½ �Fþð1-fFÞ TiO2½ �M

; ð3Þ

where d49Ti is the Ti isotopic composition expressed as thedeviation in permil of the 49Ti/47Ti ratio relative to theOL-Ti standard. The weight ratios Th/Sc and La/Sc (bothppm/ppm) are also good indicators of the nature of thesediment provenance (Taylor and McLennan 1985). Thus,we also use the published estimates of the average La/Sc(2.9 ± 0.5; 95% c.i.) and Th/Sc (1.0 ± 0.1; 95% c.i.) ratiosof shales of Phanerozoic age (0.6 to 0.0 Gyr) from Taylorand McLennan (1985), to test if our model results areconsistent between ratios normalized to Sc and ratiosnormalized to TiO2. For those ratios, the mass-balanceequations take the form:

La

Sc

� �shale

¼ fF Sc½ �F LaSc

� �Fþð1-fFÞ Sc½ �M La

Sc

� �M

fF Sc½ �Fþð1-fFÞ Sc½ �M; ð4Þ

Th

Sc

� �shale

¼ fF Sc½ �F ThSc

� �Fþð1-fFÞ Sc½ �M Th

Sc

� �M

fF Sc½ �Fþð1-fFÞ Sc½ �M; ð5Þ

Table 1Al2O3, TiO2, Zr, Th, Sc and La concentrations in mafic arc-rocks, ocean island basalts, continental flood basalts, mid ocean ridge basalts and intra-continental basalts.

Al2O3

(wt%)2SE n TiO2

(wt%)2SE n Zr

(mg/g)2SE n Th

(mg/g)2SE n Sc

(mg/g)2SE n La

(mg/g)2SE n Al2O3/

TiO22SE Source

Arc-basalts

Arc basalts, PetDB 17.2 0.3 441 0.88 0.03 441 56 3 380 0.9 0.1 298 34.6 1.1 278 7.0 0.6 292 19.6 0.79 PetDBContinental Arcbasalts, Kelemen

15.7 497 0.98 497 93 145 2.0 111 32.5 112 11.9 168 16.0 Kelemen et al.(2014)

Oceanic Arc basalts,Kelemen

15.7 503 0.91 503 62 54 1.5 34 36.4 45 7.0 59 17.3 Kelemen et al.(2014)

Ocean Island basalts 13.3 0.2 1563 3.09 0.16 1563 176 8 706 2.3 0.2 453 45.5 3.7 593 19.2 1.4 502 4.3 0.22 PetDB

Iceland 14.1 0.0 3624 2.03 0.03 3623 109.6 4.3 1069 1.0 0.1 573 10.2 0.6 748 40.5 0.5 814 6.9 0.10 GeoRoc

Continental flood basalt

provinces

Camp 15.2 0.2 343 0.96 0.07 343 81 4 277 1.5 0.1 283 37.9 0.7 281 8.4 0.6 269 15.8 1.15 GeoRocEmeishan 13.4 0.4 314 2.61 0.13 314 184 13 273 3.5 0.2 274 31.0 1.2 169 24.2 1.6 278 5.1 0.30 GeoRocMac Kenzie 13.2 0.3 240 1.68 0.12 240 171 23 224 2.2 0.6 33 55.5 4.5 116 18.4 3.3 61 7.9 0.58 GeoRocDeccan Traps 14.0 0.1 1536 2.33 0.04 1530 168 5 1212 3.2 0.4 277 34.6 0.7 719 23.4 2.6 413 6.0 0.10 GeoRocKaroo-Ferrar 14.4 0.2 288 1.58 0.11 288 130 11 279 1.7 0.5 110 32.7 0.7 173 12.8 1.6 147 9.2 0.67 GeoRocSiberian Traps 13.7 0.4 406 1.37 0.11 406 118 12 270 1.8 0.3 239 41.7 7.1 253 14.7 2.1 260 10.0 0.83 GeoRocFranklin 14.1 0.6 21 1.31 0.14 21 77 4 21 1.2 0.2 20 26.5 8.7 19 6.7 0.6 20 10.7 1.25 GeoRocParana 14.2 0.1 366 2.35 0.14 366 184 9 333 3.3 0.2 164 35.8 0.9 109 24.2 1.6 252 6.0 0.36 GeoRocColumbia River Basalt 14.3 0.1 446 2.70 0.07 446 186 6 446 3.7 0.2 252 37.3 0.5 446 24.2 1.7 255 5.3 0.14 Hooper (2000)

Intra-continental

basalts

Chaıne des Puys,France

17.0 0.5 9 2.29 0.13 9 253 30 9 6.6 0.8 9 21.0 4.1 9.0 52.7 5.6 9 7.4 0.48 Hamelin et al.(2009)

Duofutun, centralChina

14.9 0.4 8 2.36 0.14 8 248 26 8 2.6 0.2 8 24.0 0.9 8 24.5 2.1 8 6.3 0.40 Li et al. (2015)

Mid ocean ridge basalts 14.7 0.1 430 1.68 0.05 430 117 8 412 0.4 0.1 395 39.8 0.8 338 5.2 0.5 412 8.8 0.27 Gale et al.(2013)

Unfiltered data from GeoRoc and PetDB databases are available in supplementary Table S4.

254N.D

.Greb

er,N.Dau

phas

/Geochim

icaet

Cosm

ochim

icaActa

255(2019)

247–264


As can be seen in Eqs. (1)–(5), next to the compositionof the fine-grained terrigenous sediments, the most impor-tant input parameters for our model are the concentrationsand elemental ratios of the rock endmembers. A potentialdifficulty in the mass-balance approach in the Phanerozoicis to estimate the composition of the mafic endmember, asmafic rocks from Arc-settings (Arc-basalts) are depleted insome elements such as Ti and Zr compared to within platebasalts (WPB) that include Ocean Island basalts (OIB),continental flood basalts provinces (LIPs) and intra-continental basalts not associated with LIPs (Fig. 5 andTable 1). A compilation of whole rock compositions ofWPB and Arc-basalt samples shows that the Al2O3/TiO2

ratios are variable, ranging from low values in OIB andLIP basalts (e.g., Emeishan LIP = 5.1 ± 0.3; 2SE) to highvalues in Arc-basalts (19.6 ± 0.8; 2SE). The Zr/TiO2 ratiois more or less constant among these different mafic rocks(Table 1). The depletion of Arc-basalts in Ti and high-field-strength elements in general (HFSE; e.g., Hf, Zr, Ti,Nb, Ta) relative to mid-ocean ridge basalts is well-documented and has been attributed to the presence ofHFSE-rich minerals like rutile or ilmenite in the subductionzone system or to the lower mobility of HFSE in fluidsresponsible for the metasomatism of the depleted mantlewedge (Woodhead et al., 1998; Ulmer, 2001; Munkeret al., 2004; Kelemen et al., 2014; 1990). Furthermore, itwas also suggested that the high Ti concentration in OIBscannot be achieved solely by melting of a peridotitic mantle,but instead requires a small amount of recycled mafic crustin its source (Prytulak and Elliott, 2007). To account forthis chemical diversity of modern mafic magmatic rocks,we split the mafic endmember into WPB and Arc-basalts.We note fWPB ¼ mWPB=ðmWPB þ mArc þ mF Þ andf Arc ¼ 1� f WPB � f F , the proportions of these endmembersin the sediment provenance. Data from the Emeishan LIPhas been used for the Al2O3, TiO2, Zr, La, Th and Sc con-centrations of the WPB endmember. The reasoning behindthis is that OIB and non-LIP intra-continental basalts areassumed to be volumetrically less important for thePhanerozoic emerged crust compared to LIP and Arc-basalts. Figure 5 shows that mixing between the EmeishanLIP and the calculated average Arc-basalt composition ofthe PetDB database can explain the composition of allother types of basalts for the ratios that we are interestedin (black dotted line in Fig. 5), which is the reason whythe composition of the Emeishan LIP is used for the LIPendmember.

One can obtain exact solutions for the proportions of allrock types in the emerged crust (3 unknowns; fF, fArc, fWPB)by writing sets of mass-balance equations (e.g., Eqs. (6)–(7)–(11), (6)–(8)–(11), (6)–(9)–(11) or (6)–(10)–(11)), allinvolving the Al2O3/TiO2 ratio, which distinguishesbetween WPB and Arc-basalts (in addition to distinguish-ing between mafic and felsic rocks), and either the Zr/TiO2, Th/Sc, La/Sc ratios or d

49Ti value, which distinguishbetween mafic and felsic rocks:

Al2O3

TiO2

� �shale

¼fF TiO2½ �F Al2O3

TiO2

� �FþfArc TiO2½ �Arc

Al2O3

TiO2

� �Arc

þfWPB TiO2½ �WPBAl2O3

TiO2

� �WPB

fF TiO2½ �FþfArc TiO2½ �ArcþfWPB TiO2½ �WPB

;

ð6Þ

Zr

TiO2

� �shale

¼fF TiO2½ �F Zr

TiO2


ZrTiO2

� �Arc

þfWPB TiO2½ �WPBZr

TiO2

� �WPB


ð7ÞTh

Sc

� �shale

¼fF Sc½ �F ThSc

� �FþfArc Sc½ �Arc

ThSc

� �Arc

þfWPB Sc½ �WPBThSc

� �WPB

fF Sc½ �FþfArc Sc½ �ArcþfWPB Sc½ �WPB

ð8Þ

La

Sc

� �shale

¼ fF Sc½ �F LaSc


LaSc

� �Arc

þfWPB TiO2½ �WPBLaSc

� �WPB

fF Sc½ �FþfArc Sc½ �ArcþfWPB Sc½ �WPB

ð9Þ

d49Tishale¼fF TiO2½ �Fd49TiFþfArc TiO2½ �Arcd49TiArcþfWPB TiO2½ �WPBd

49TiWPB


;

ð10Þ1 ¼ fFþfArcþfWPB: ð11Þ

All the compositions of the rock endmembers and thefine-grained terrigenous sediments needed to solve Eqs.(6)–(10) are summarized in supplementary Table S2. Theresults of the various sets of mass-equations are given inTable 2 and shown in Fig. 6. The errors on these estimatesare 95% confidence intervals and have been calculated fol-lowing the methodology outlined in Greber et al. (2017a).The solutions involving the various pairs of proxy ratiosall agree within error, indicating that felsic material repre-sented between 67 ± 7 and 84 ± 15 wt% of the emergedcrust 250 Myr ago. As expected, the Al2O3/TiO2 - Zr/TiO2 pair results in the lowest estimate of felsic materialbecause zircon grains can be scavenged into sandstones.The combination of Al2O3/TiO2 and d49Ti yields the high-est proportion of felsic rocks in the emerged crust 250 Myrago and the rather large error on this estimate is due to thequasi parallel evolution of these two equations at a high fel-sic rock fraction (Fig. 6). A small change in the d49Ti valueof the WPB endmember from +0.005‰ (the average valueof basalts from diverse tectonic settings; Millet et al., 2016)to +0.04 ‰ (a plausible value for tholeiitic basalts; Denget al., 2019), would shift the Ti isotope curve in Fig. 6 sothat it overlaps with the solutions given by the La/Sc andTh/Sc ratios. Averaging the results of the 5 different sys-tems (Al2O3/TiO2, Zr/TiO2, La/Sc, Th/Sc and Ti isotopes)without further corrections indicates that the emerged con-tinents 250 Myr ago consisted of 76 ± 8 wt% felsic rocks,14 ± 6 wt% Arc-basalts and 10 ± 2 wt% within platebasalts.

This result can be compared to independent estimates ofthe felsic/mafic rock ratio in the emerged continents basedon surface mapping (supplementary Table S3 and graybar in Fig. 6) (Condie, 1993; Wedepohl, 1995; Durr et al.,2005; Moosdorf et al., 2010; Hartmann and Moosdorf,2012). Examining mapped mafic and felsic rocks, Condie(1993) and Wedepohl (1995) suggested that the Phanerozoicemerged continents contained 87 and 89 wt% felsic litholo-gies, respectively. Digital lithological maps of north Amer-ica (Moosdorf et al., 2010) and Earth as a whole (Durret al., 2005; Hartmann and Moosdorf, 2012) suggest alower proportion of felsic rocks ranging from 61 to 77 wt% (supplementary Table S3). Overall, the felsic/mafic rockratio in the emerged crust from mapping agrees with ourestimate based on the sediment record and shows that ourapproach does not suffer from any obvious bias and canbe applied to test whether the Paleoarchean emerged crustwas dominated by mafic lithologies or not.

Table 2Summary of results of our endmember modelling.

Proxies Felsic 95c.i. MaficA 95c.i Komatiite 95c.i. Arc-basalts 95c.i. WPB 95c.i.

Modern crust, 250 Myr

Al2O3/TiO2 and d49Ti 84 15 16 15 – – 5 5 11 3.2Th/Sc and Al2O3/TiO2 78 3 22 3 – – 12 4 10 1.2La/Sc and Al2O3/TiO2 75 6 25 6 – – 15 7 10 1.5Zr/TiO2 and Al2O3/TiO2 67 7 33 7 – – 25 9 8 2.0

Average 250 Myr 76 8 24 8 – – 14 6 10 2

Paleoarchen crust, 3250 Myr

Al2O3/TiO2 and Ni/Co 64 5 25 5 12 4 – – – –Al2O3/TiO2 and Cr/Sc 65 5 25 5 10 3 – – – –Zr/TiO2 and Ni/Co 64 7 24 7 12 4 – – – –Zr/TiO2 and Cr/Sc 65 7 25 6 10 3 – – – –La/Sc and Ni/Co 66 8 23 7 11 4 – – – –La/Sc and Cr/Sc 66 8 24 7 10 3 – – – –Th/Sc and Ni/Co 68 7 21 6 11 3 – – – –Th/Sc and Cr/Sc 68 7 22 6 10 3 – – – –d49Ti and Ni/Co 60 8 28 7 12 4 – – – –d49Ti and Cr/Sc 60 8 29 7 11 3 – – – –

Average Paleoarchean 65 7 25 6 11 3 – – – –

A: Mafic for modern crust is total of Arc-basalts and WPB.WPB: within plate basalts.

256N.D

.Greb

er,N.Dau

phas

/Geochim

icaet

Cosm

ochim

icaActa

255(2019)

247–264

Fig. 6. Calculated proportions of felsic rocks in the emerged crust(x-axis) relative to the ratio of within plate basalts to Arc-basalts(y-axis) that explain the Al2O3/TiO2 (blue), Zr/TiO2 (green), La/Sc(olive), Th/Sc (red) and d49Ti (orange) values of the Phanerozoicfine-grained terrigenous sediment record (see supplementaryTable S2). The grey bar with black dotted line is the mean(82 ± 7 wt%; 2SE, n = 4) of different estimates of the abundance offelsic rocks in the modern emerged continents based on surfacemaps (see supplementary Table S3). (For interpretation of thereferences to colour in this figure legend, the reader is referred tothe web version of this article.)


To the best of our knowledge, it is the first time that thecontributions of Arc-basalts and WPBs to fine-grained ter-rigenous sediments are estimated. A caveat, however, is thatwe chose the Emeishan LIP as WPB endmember due to itslow Al2O3/TiO2 ratio. Therefore, the contribution of WPBto the sediment flux is likely underestimated. As shown inFig. 6, our estimate of the felsic and mafic proportionsare robust, as the proxy ratio reconstructions show littlesensitivity to the abundance of felsic rocks when the ratioWPB/Arc-basalts is higher than �0.5.

4.4. Reconstructing the lithological composition of the

Paleoarchean emerged crust

The rocks found in the Paleoarchean crust are of differ-ent nature than those found more recently, calling for acareful assessment of the rock endmembers used in the mix-ing model. Mafic rocks of Archean age are more homoge-neous in their Ti and Zr concentrations than those in thepost-Archean (Moyen and Laurent, 2018), so there is noneed to divide the mafic endmember into sub-categories.Another distinguishing feature of the Archean crust is thecommon occurrence of ultramafic komatiitic magmas(Arndt, 2003; Condie and O’Neill, 2011), lavas with aunique chemical composition characterized by unusuallyhigh Mg, Ni and Cr concentrations. Consequently, toreconstruct the lithological composition of the emergedcrust 3.25 Gyr ago, we rewrite our endmember mixingmodel to quantify the relative abundances of felsic, maficand komatiitic (fK) lithologies. For these calculations, weuse (i) the Al2O3/TiO2 (wt%/wt%), Zr/TiO2 (ppm/wt%),La/Sc (ppm/ppm) and Th/Sc (ppm/ppm) weight ratios, as

well as d49Ti values due to their sensitivities to discriminatebetween Archean felsic and mafic rocks (Fig. 1 and Greberet al., 2017a) and (ii) the Ni/Co and Cr/Sc ratios as indica-tors of the contribution of komatiites (Greber et al., 2017a).The following mixing equations can be written:

Al2O3

TiO2

� �shale

¼fF TiO2½ �F-A Al2O3

TiO2

� �F-A

þfM TiO2½ �M-AAl2O3

TiO2

� �M-A

þfK TiO2½ �K Al2O3

TiO2

� �K

fF TiO2½ �F-AþfM TiO2½ �M-AþfK TiO2½ �K;

ð12Þ

Zr

TiO2

� �shale

¼fF TiO2½ �F-A Zr

TiO2

� �F-A

þfM TiO2½ �M-AZr

TiO2

� �M-A

þfK TiO2½ �K ZrTiO2

� �K

fF TiO2½ �F-AþfM TiO2½ �M-AþfK TiO2½ �K;

ð13Þ

d49Tishale ¼ fF TiO2½ �F-Ad49TiFþfM TiO2½ �M-Ad49TiMþfK TiO2½ �Kd49TiK

fF TiO2½ �FþfM TiO2½ �M-AþfK TiO2½ �K; ð14Þ

Th

Sc

� �shale

¼fF Sc½ �F-A ThSc

� �F-A

þfM-A Sc½ �M-AThSc

� �M-A

þfK Sc½ �K ThSc

� �K

fF-A Sc½ �F-AþfM-A Sc½ �M-AþfK Sc½ �Kð15Þ

La

Sc

� �shale

¼ fF Sc½ �F-A LaSc

� �F-A

þfM-A Sc½ �M-ALaSc

� �M-A

þfK Sc½ �K LaSc

� �K

fF-A Sc½ �F-AþfM-A Sc½ �M-AþfK Sc½ �Kð16Þ

Ni

Co

� �shale

¼ fF Co½ �F-A NiCo

� �F-A

þfM-A Co½ �M-ANiCo

� �M-A

þfK Co½ �K NiCo

� �K

fF-A Co½ �F-AþfM-A Co½ �M-AþfK Co½ �K; ð17Þ

Cr

Sc

� �shale

¼fF Sc½ �F-A CrSc

� �F-A

þfM-A Sc½ �M-ACrSc

� �M-A

þfK Sc½ �K CrSc

� �K

fF-A Sc½ �F-AþfM-A Sc½ �M-AþfK Sc½ �K; ð18Þ

1 ¼fFþfMþfK: ð19ÞHere, subscripts F-A and M-A indicate the Archean fel-

sic and mafic endmember compositions, which are differentfrom the modern (Keller and Schoene, 2018; 2012). All theparameters needed to solve Eqs. (12)–(19) are presented insupplementary Table S2. The Ni/Co and Cr/Sc ratios offine-grained sediments of various ages are from the datacompilation of Greber et al. (2017a) and we only consideredthe ratios that passed the filter to remove samples obviouslyaffected by weathering. The samples that did not pass thefilter of Greber et al. (2017a) are indicate in ‘‘red” in thesupplementary Table S1. For consistency with the sedimen-tary Al2O3/TiO2 and Zr/TiO2 records, Ni/Co and Cr/Scratios of sediments were also averaged by location (seeFig. 3C and 3D).

We can calculate the proportion of felsic rocks needed inthe emerged crust to explain the Al2O3/TiO2, Zr/TiO2,

La/Sc, Th/Sc, d49Ti, Ni/Co and Cr/Sc signatures of thefine-grained sediments 3.25 Gyr ago, leaving the komatiiticto mafic rock ratio as a free parameter. This is illustrated inFig. 7, where the felsic rock proportion (f FÞ of the emergedcrust is plotted on the x-axis and the ratio of komatiite tomafic fractions (fK=fM) is plotted on the y-axis. The resultsreveal that the solutions for Al2O3/TiO2, Zr/TiO2, La/Sc,Th/Sc and d49Ti plot in a very narrow range and followsteep, subparallel curves. The solution for the Zr/TiO2 sig-nature of the sediments plots within the range defined bythe other proxies and unlike the modern, is not shiftedtowards a lower proportion of felsic rocks. This agrees withour previous observation that the average Zr/Al2O3 ratiosof Archean sediments and igneous rocks are within erroridentical, which suggests that mineral sorting processes dur-ing the Archean were limited, at least for the set of terrige-nous sediments investigated here. As discussed previously,one factor that controls the degree of zircon sorting intosandstones is the transport regime and transport distance


that the sediment grains are subjected to (Garconet al.,2013b). Thus, an explanation for the limited mineral sortingmight be that the areal extent of exposed landmasses wassmall in the Paleoarchean (Rey and Coltice, 2008;Flament et al., 2013; Bindeman et al., 2018). Alternatively,zircon saturation in rocks of Paleoarchean age was likelyachieved less frequently 3.25 Gyr ago than in the moderndue to higher degrees of partial melting and lower concen-trations in incompatible element in mantle-derived magmas(Keller et al., 2017), a matter that would have limited theeffect of hydrodynamic grain size sorting on the Zr concen-tration of fine-grained terrigenous sediments.

As can be seen in Fig. 7, the solution space defined byCr/Sc and Ni/Co ratios is very different from that definedby the other proxies. The required amount of felsic rocksto successfully solve Eqs. (17) and (18) and to explain theCr/Sc and Ni/Co ratios of terrigenous sediments stronglydepends on the ratio of komatiitic to mafic rocks in theemerged crust. The solutions of all 7 equations intersectin a small window. The composition of the fine-grainedterrigenous sediment record 3.25 Gyr ago can only beexplained with an emerged crust that contained 65 ± 7 wt%felsic, 25 ± 6 wt% mafic and 11 ± 3 wt% komatiitic materi-als (Table 2). This estimate also broadly agrees with thereconstructed composition of the Early Archean emergedcrust based on the currently exposed igneous rock recordof �75 wt% felsic rocks and a komatiite to mafic rock ratioof 0.4 (Condie, 1993) (yellow star in Fig. 7).

We tested if a shift in the dominant type of komatiiticrocks (Al-depleted, Al-undepleted, Al-enriched) from theearly to the late Archean impacts our model results, bychanging the composition of the komatiite endmember.To do so, we used the chemical compositions of the chilledmargins from the Komati formation (Al-depleted) and Wel-tevreden formation (Al-enriched) komatiites from Puchtelet al. (2013). As can be seen in supplementary Fig. S6 thenature of the komatiite endmember has no significant influ-ence on the result.

Deng et al., (2019) suggested that the Ti isotopic compo-sition of Archean sediments might not be the result of theerosion of a felsic emerged crust, but of continents madewith dominantly tholeiitic rocks. We tested this possibilityby redefining the element concentrations of our felsic andmafic endmembers using only tholeiitic rocks of Archeanage (see supplementary Table S2 and Fig. S1). As can beseen in Fig. S7, applying this new endmember classification,the solutions for the element ratios Al2O3/TiO2, Zr/TiO2,Th/Sc, and La/Sc show a larger scatter than in our initialmodel, but still call for a high percentage of igneous felsicmaterial in the Paleoarchean emerged crust (�55wt%). Thisvalue agrees within error with the estimate based the Ti iso-tope proxy of 60 ± 8 wt% (see Table 2 and Greber et al.,2017a) and indicates that either tholeiitic magmatic seriesin the Archean did not produce a significantly different Tiisotope vs. SiO2 trend than calc-alkaline magmatism, orthat the Paleoarchean continents consisted mainly of calc-alkaline rocks, especially among more evolved units. Thelatter is in agreement with a recent study, suggesting thatsince 3.8 Gyr calc-alkaline differentiation trends dominatedcontinental magmatism (Keller and Schoene 2018). There-

fore, we argue that considering the whole Archean rock cat-alogue leads to the most reliable definition of the variousArchean rock endmembers, and translates the sedimentrecord into the lithological composition of the emergedcrust most accurately.

At first sight, it might be even counter intuitive that ahigher Al2O3/TiO2 ratio of fine-grained terrigenous sedi-ments in the Paleoarchean (�27.7) compared to thePhanerozoic (�22.0) translates into a lower estimate forthe proportion of felsic material in the emerged crust. A fac-tor explaining the decreasing trend in the sedimentaryAl2O3/TiO2 ratio with time (see Fig. 3A) is the significantincrease in the TiO2 concentration of mafic rocks overEarth’s history reflecting lower degrees of melting and asso-ciated higher concentrations of highly incompatible ele-ments in more recent mafic magmas (Keller and Schoene,2018). Because Al2O3 is overall more compatible thanTiO2 during mantle melting processes (but not during mag-matic differentiation as shown by the increasing Al2O3/TiO2

with increasing SiO2 content; Fig. 1), lower degrees of par-tial melting shifts the average Al2O3/TiO2 ratio of the conti-nents towards lower values, despite the fact that the amountof mafic rocks did not vary much. The Zr concentration inmafic rocks increases in a similar way as the TiO2 concentra-tion (Keller and Schoene, 2018), which explains why the Zr/TiO2 ratio in fine-grained terrigenous sediments does notexhibit the same secular trend as the Al2O3/TiO2 ratio.

The results presented here show that the composition ofPaleoarchean fine-grained terrigenous sediments requiresan emerged crust with �65 wt% felsic material (Table 2).The Paleoarchean emerged continents were more maficand ultramafic compared to their modern counterparts,but to a lesser extent than was suggested previously(Dhuime et al., 2015; Tang et al., 2016; Large et al.,2018). Consequently, the high Ni and Cr concentrationsand accompanying high Ni/Co, Cr/Zn, Cr/U, Cr/Th andCr/La ratios in 3.25 Gyr old sediments can no longer betaken as evidence for an emerged crust dominated by maficrocks. Figure 7 indeed reveals that the Ni/Co and Cr/Scratios of 3.25 Gyr old fine-grained terrigenous sedimentscan be explained by an emerged crust that contains 0 to90 wt% felsic material, depending on the amount of komati-itic rocks present. This means, that the concentrations of Niand Cr in Paleoarchean sediments are largely controlled bythe abundance of ultramafic rocks and komatiites, so theseelements are not sensitive proxies to reconstruct the propor-tion of felsic material in the emerged crust during that timeperiod.

4.5. Regional diversity in the lithological makeup of Archean

continents

The new dataset of terrigenous sediment compositionsallows us to investigate whether regional variations existedin the lithological composition of the Archean continents.To do so, we subdivided the Archean sediments of our aver-aged shale database into samples from (1) Southern Africa(Zimbabwe, South Africa), (2) North America – Greenland,(3) Australia, (4) India and (5) Ukraine, and plotted them inthe Cr/Sc – Al2O3/TiO2 diagram (Fig. 8). The sediment

Fig. 7. Calculated proportion of felsic rocks in the emerged crust(x-axis) relative to the weight ratio of komatiites to mafic rocks (y-axis) that explains the Al2O3/TiO2 (blue), Zr/TiO2 (green), La/Sc(olive), Th/Sc (red), d49Ti (orange), Ni/Co (black) and Cr/Sc(brown) values of the Paleoarchean fine-grained terrigenoussediment record (Al2O3/TiO2 = 27.7 ± 1.4; Zr/TiO2 = 245 ± 32;La/Sc = 1.4 ± 0.3; Th/Sc = 0.46 ± 0.09; Ni/Co = 6.9 ± 0.9; Cr/Sc = 26.7 ± 4.0 and d49Ti = 0.160 ± 0.023‰, see supplementaryTable S2). All seven proxies intersect in a common narrow window(grey box), indicating that the emerged crust 3.25 Gyr agocontained approximately 65 wt% felsic material. The yellow staris the estimate from Condi (1993) based on mapping Early Archeanterranes. (For interpretation of the references to colour in thisfigure legend, the reader is referred to the web version of thisarticle.)

Fig. 8. Cr/Sc vs. Al2O3/TiO2 weight ratios of location averagedsamples of Archean age (3.8 to 2.5 Gyr), color-coded according totheir origin and shape-coded according to their age. Indicated incoarse lines is the mixing space defined by the felsic, mafic andkomatiitic endmembers, with crosses marking each 10% mixingstep. Small dotted lines correspond to a provenance with 60 wt%felsic and 10 wt% komatiitic components. Filled diamonds are themean values of each location. The star is the result of ourendmember model inversion for the emerged crust 3.25 Gyr ago.


data are encompassed by the ternary mixing space definedby the felsic, mafic and komatiite endmembers. There isno obvious trend towards a more felsic composition ofthe emerged continents from the Paleoarchean (squares)to the Meso- and Neoarchean (circles). As seen from thisdiagram, the broad secular trend in the chemical composi-tion of terrigenous sediments used to reconstruct the natureof the continents through time (this study; Tang et al., 2016;Greber et al., 2017a; Smit and Mezger 2017) hides a lot ofregional diversity. While the data points define an averageArchean crust comprising around 60% felsic, 30% mafic,and 10% komatiite rocks, the contributions of theseendmembers to individual shale localities shows a lot ofvariation with komatiite contributions that vary between�0 and 50% and felsic fractions between 0 and 90%. Theredoes not seem to be any systematics from craton to cratonin the proportions of felsic to mafic rocks. The variationsseem to be primarily driven by the random distribution ofmafic and felsic rocks in a given drainage basin. There ishowever some systematic craton-to-craton variation in thefraction of komatiites. Sediments sampled in the SouthAfrica craton incorporated a larger fraction of komatiitesthan average, while the Indian craton is characterized bya lower komatiite fraction than average. This finding is cor-roborated by the observation of Taylor and McLennan(1985) that the Archean greenstone belts from Barberton(South Africa) and from the Yilgarn Block (Australia)contain the highest abundance of komatiitic material.

4.6. Insights into Paleoarchean geodynamics

The question of the prevailing mode of tectonism duringdifferent periods of Earth’s history has long been debated(see the following papers for an overview of the presentstate of this debate: Cawood et al., 2018; Foley, 2018;Hawkesworth and Brown, 2018; Korenaga, 2018;Lenardic, 2018). It is generally accepted that magmatismand other associated processes in subduction zone systemsproduce felsic lithologies (Barbarin, 1999; Jagoutz et al.,2009; Jagoutz, 2010; Grove et al., 2012; Moyen andMartin, 2012). In modern settings, the transport of waterinto the mantle via subduction zones can lead to:

(i) Metasomatism and melting of the mantle wedgethrough dehydration of the slab, producing basalticmelts. These melts are then modified through frac-tional crystallization, mixing, and assimilation duringtheir ascent towards the surface, resulting in the pro-duction of felsic lithologies (Hildreth and Moorbath,1988; Annen et al., 2005; Muntener and Ulmer,2018). This leads to the segregation between shallowfelsic and deep mafic lithologies. To finally produce abulk continental crust of andesitic composition, addi-tional processes such as delamination of lower crustmafic litholgies need to be at work (Rudnick andGao, 2003; Jagoutz and Kelemen, 2015).


(ii) In hot subduction settings, the hot and hydrated sub-ducting slab itself can melt, producing rare felsicrocks named adakites (Defant and Drummond,1990). A similar mechanism involving partial meltingof hydrated subducted slabs in the presence ofamphibole and/or garnet may have produced felsicArchean TTGs (Foley et al., 2002; Moyen andMartin, 2012).

In either case (i or ii), significant volumes of felsic rocksare expected to be produced in a subduction zone setting.Thus, a prominent line of argument for the absence of platetectonics on Earth until 3.0 Gyr is based on the claim thatthe continental crust until that time was dominated bymafic rocks (Dhuime et al., 2015; Tang et al., 2016).

The composition of the terrigenous sediments has notbeen the only line of evidence used to argue for predomi-nantly mafic continents in the Paleoarchean. Dhuimeet al., (2015) used the Sm-Nd and Sr isotopic compositionsof igneous rocks to estimate the average Rb/Sr ratio of the‘‘juvenile continental crust” (the composition of mantle-derived magma that crystallized in the crust after differenti-ation). For a given igneous rock of a given age and initial143Nd/144Nd and 87Sr/86Sr ratios, they calculate a modelage of crustal extraction from a depleted mantle sourceusing 147Sm-143Nd systematics. They then calculate theRb/Sr ratio needed to bring the 87Sr/86Sr ratio from adepleted mantle value at the time of extraction (modelage given by 147Sm-143Nd) to the initial value measured atthe time of crystallization. As the Rb/Sr of igneous rockscorrelates positively with their SiO2 concentration, theincrease in the calculated Rb/Sr ratio of juvenile continen-tal crust from 3 to 2 Gyr was taken as evidence that thecomposition of the continental crust changed from maficto andesitic over that time period (Dhuime et al., 2015).The findings of a bulk mafic continental crust (SiO2

�49 wt%) prior to 3.0 Gyr is at odds with the currentlyexposed rock record of that time (Condie, 1993). A solutionto this problem could be, that large parts of the maficlithologies were eroded and are no longer present. How-ever, as shown by Greber et al. (2017a) and the presentstudy, such a process is not recorded in terrigenous sedi-ments. There are two potential issues with the reconstruc-tion of Dhuime et al. (2015) based on 87Rb-86Sr and147Sm-143Nd systematics of igneous rocks:

(i) As seen in Fig. 9A, the inferred composition of thecontinents for the time prior to 3.0 Gyr (i.e., sampleswith mantel extraction model ages TDM � 3.0 Gyr)relies mostly on rocks that crystallized within the last500 Myr (i.e., crystallization age tCA � 500 Myr).Almost all of them have low modelled Rb/Sr ratios.Thus, for the model to work, one has to assume thatthe Sm-Nd and Rb-Sr systematics evolved as closedsystems and remained undisturbed over billions ofyears despite magmatic and/or metamorphicreworking.

(ii) The calculated Sm-Nd mantle extraction model agesassume that all rocks were derived from a depletedmantle reservoir. The depleted mantle, however, is

an extreme endmember that can bias the mantleextraction ages towards higher values. For example,in the study of Dhuime et al. (2015) rocks from theDeccan Traps flood basalts (�65 Myr)(Vanderkluysen et al., 2011) and basalts from theKerguelen Archipelago (�25 Myr) (Doucet et al.,2005) were used. In both cases, the ages of theserocks should be similar to their mantle extractionages. However, applying the mantle extraction agecalculation of Dhuime et al. (2015) results in TDM

ranging from 819 to 1190 Myr for samples from theKerguelen system and from 1010 to 4366 Myr forsamples from the Deccan Trap basalts. In both cases,their mantle extraction ages are overestimated, whichleads to a gross underestimation of the Rb/Sr ratio ofthe juvenile continental crust. Indeed, an older man-tle extraction age will mean more time to produceradiogenic 87Sr so a lower Rb/Sr ratio in the juvenilecontinental crust is called for to explain a given initial87Sr/86Sr ratio at crystallization. This issue is illus-trated in Fig. 9B, where the modelled Rb/Sr ratio isplotted as a function of the crystallization age ofthe rocks with a TDM of � 3.0 Gyr, i.e. data that isused to constrain the Rb/Sr ratio of juvenile conti-nental crust added to the continents before 3.0 Gyr(Dhuime et al. 2015). As can be seen, rocks that crys-tallized within the last 500 Myr yield lower modelledmedian Rb/Sr ratios than rocks of older age. Themedian Rb/Sr model ratio of rocks that crystallizedin the Archean (tCA � 2.5 Gyr) and are thus likelythe samples least affected by the problems outlinedabove is 0.088. This ratio translates into SiO2 concen-tration of �57 wt%, close to the average SiO2 concen-tration of the modern bulk continental crust ofRudnick and Gao (2003) of 60.6 wt%.

Based on the combined use of Ni/Co and d49Ti measure-ments of fine-grained terrigenous sediments, we previouslyconcluded that the crust was predominantly felsic in theArchean. The present study shows that available concentra-tion data for Al, Ti, Zr, Th, La, Ni, Co, Cr and Sc in fine-grained terrigenous sediments corroborate this conclusion.It also agrees with the inventory of exposed Early Archeanigneous rocks (Condie, 1993). Following the argument ofDhuime et al. (2015) and Tang et al. (2016), our findingsof a predominantly felsic crust since 3.25 Gyr would sup-port the view that subduction-zones were already active atthat time.

The argument has been made, however, that chemicallyevolved continents could be formed without subductionzones. The Archean rocks of the Pilbara Craton in Aus-tralia are often put forward as an example of how felsicrocks in the Archean were produced in thick volcanic pla-teau complexes (Van Kranendonk et al., 2015). However,it is unclear if these settings are capable of producing anemerged continental crust that contained on average65 wt% of felsic material. A petrological–thermomechanicalnumerical model investigating the formation of continentalcrust via drip tectonics estimated a ratio of felsic to maficrocks in granite-greenstone terranes of 1.2 (Sizova et al.,

Fig. 9. Reevaluation of the data used in Dhuime et al. (2015). (A) Mantle extraction age based on Sm-Nd isotope systematic vs. crystallizationage of the rocks, color coded after calculated juvenile Rb/Sr ratios. A large fraction of the low Rb/Sr weight ratios for the pre-3.0 Gyr juvenilecontinental crust derives from rocks that apparently resided in the continental crust for more than 2.5 Ga. We argue that the very old mantleextraction ages (>3.0 Gyr) of young rocks (<500 Myr) are overestimates resulting from the assumption that their mantle source was depleted.(B) Modelled Rb/Sr ratio of juvenile continental crust vs. crystallization age of the rocks with a mantle extraction age of � 3.0 Gyr. Rockswith a young crystallization age have lower modelled Rb/Sr ratios than rocks with an old crystallization age. The red bar is the median Rb/Srratio (0.088) of rocks with a crystallization age of above 2.5 Gyr. (For interpretation of the references to colour in this figure legend, the readeris referred to the web version of this article.)


2015). This is a factor of 2 lower than the ratio of 2.6 esti-mated in this study that represents the composition of theemerged continents, and a factor of 3.5 lower than the ratioobserved in currently exposed Early Archean terranes(Condie, 1993). The latter value might be more representa-tive of deeper parts of Early Archean continents due to theeffect of erosion.

To summarize, the lithological and chemical compositionof the Paleoarchean emerged crust should no longer be usedas argument against the existence of subduction zones at thattime. Indeed, the geochemistry of terrigenous sediments ofPaleoarchean age points to exposed rocks at the surface ofcontinents that were predominantly felsic; supporting theview that some form of subduction processes were operatingsince the early Archean (Hopkins et al., 2010; Polat et al.,2015; Smart et al., 2016; Reimink et al., 2018).

5. CONCLUSIONS

A new compilation of Al2O3, TiO2 and Zr concentra-tions of fine-grained terrigenous sediments was used toquantify the rock composition of the emerged crust3.25 Gyr ago. We show that the composition of detrital sed-iments in the Paleoarchean is best explained with thepresence of �65 wt% felsic rocks and �11 wt% komatiiticand/or ultramafic material in the emerged crust. From

our estimate, the proportion of mafic and ultramafic rocksdecreased from around 35 to 24 wt% from the Archean tothe Proterozoic. This shift is smaller than what was advo-cated previously and we argue that the emerged continents3.25 Gyr ago were already chemically mature, a result thatsupports the operation of some form of subduction at thattime.

We also used the Zr/Al2O3 ratio of the sediments toinvestigate whether or not hydrodynamic mineral sortinginfluenced their compositions. We find that Archean sam-ples were only marginally impacted by such processes.However, there might be a temporal correlation betweentime intervals with low Zr/Al2O3 ratios in fine-grained sed-iments and periods of supercontinents, indicating the possi-bility that large landmasses and mountain-buildingprocesses led to enhanced continental storage of dense min-erals like zircons during those times.

ACKNOWLEDGMENTS

The authors thank Matous Ptacek, Andrey Bekker, Ilya Binde-man and Josh Davies for discussions. Reviews by two anonymousreferees and Christopher Spencer substantially improved the qual-ity of the manuscript. NDG was funded by the Swiss NationalScience Foundation (project number 182007). ND was supportedby NASA grants NNX17AE86G (LARS), NNX17AE87G (Emerg-ing Worlds), and 80NSSC17K0744 (Habitable Worlds).


APPENDIX A. SUPPLEMENTARY MATERIAL

Supplementary data to this article can be found online athttps://doi.org/10.1016/j.gca.2019.04.012.

REFERENCES

Adam J. and Green T. (2006) Trace element partitioning betweenmica- and amphibole-bearing garnet lherzolite and hydrousbasanitic melt: 1. Experimental results and the investigation ofcontrols on partitioning behaviour. Contrib. Miner. Petrol. 152,1–17.

Annen C., Blundy J. D. and Sparks R. S. J. (2005) The genesis ofintermediate and silicic magmas in deep crustal hot zones. J.Petrol. 47, 505–539.

Arndt N. (2003) Komatiites, kimberlites, and boninites. J.

Geophys. Res. Solid Earth 108, 2293.Arndt N. and Davaille A. (2013) Episodic Earth evolution.

Tectonophysics 609, 661–674.Barbarin B. (1999) A review of the relationships between granitoid

types, their origins and their geodynamic environments. Lithos46, 605–626.

Bindeman I. N., Bekker A. and Zakharov O. D. (2016) Oxygenisotope perspective on crustal evolution on early Earth: Arecord of Precambrian shales with emphasis on Paleoprotero-zoic glaciations and Great Oxygenation Event. Earth Planet.

Sci. Lett. 437, 101–113.Bindeman I. N., Zakharov D. O., Palandri J., Greber N. D.,

Dauphas N., Retallack G. J., Hofmann A., Lackey J. S. andBekker A. (2018) Rapid emergence of subaerial landmasses andonset of a modern hydrologic cycle 2.5 billion years ago. Nature

557, 545–548.Bouchez J., Gaillardet J., France-Lanord C., Maurice L. and

Dutra-Maia P. (2011) Grain size control of river suspendedsediment geochemistry: Clues from Amazon River depthprofiles. Geochem. Geophys. Geosyst. 12.

Campbell I. H. and Allen C. M. (2008) Formation of supercon-tinents linked to increases in atmospheric oxygen. Nat. Geosci.

1, 554–558.Canfield D. E. (2005) The early history of atmospheric oxygen:

homage to Robert M. Garrels. Annu. Rev. Earth Planet. Sci. 33,1–36.

Cawood P. A. and Hawkesworth C. J. (2015) Temporal relationsbetween mineral deposits and global tectonic cycles London,Special Publications. Geol. Soc. 393, 9–21.

Cawood P. A., Hawkesworth C. J., Pisarevsky S. A., Dhuime B.,Capitanio F. A. and Nebel O. (2018) Geological archive of theonset of plate tectonics. Philosoph. Trans. Roy. Soc. A: Math.

Phys. Eng. Sci. 376, 20170405–20170430.Condie K. C. (1993) Chemical composition and evolution of the

upper continental crust: contrasting results from surface sam-ples and shales. Chem. Geol. 104, 1–37.

Condie K. C. (2005) Earth as an Evolving Planetary System.Academic Press.

Condie K. C. and O’Neill C. (2011) The Archean-Proterozoicboundary: 500 my of tectonic transition in Earth history. Am. J.

Sci. 310, 775–790.Condie K. C., Arndt N., Davaille A. and Puetz S. J. (2017) Zircon

age peaks: Production or preservation of continental crust?Geosphere 13, 227–234.

Defant M. J. and Drummond M. S. (1990) Derivation of somemodern arc magmas by melting of young subducted litho-sphere. Nature 347, 662–665.

Deng Z., Chaussidon M., Savage P., Robert F., Pik R. andMoynier F. (2019) Titanium isotopes as a tracer for the plume

or island arc affinity of felsic rocks. Proc. Nat. Acad. Sci. USA

116, 1132–1135.Dhuime B., Wuestefeld A. and Hawkesworth C. J. (2015)

Emergence of modern continental crust about 3 billion yearsago. Nat. Geosci. 8, 552–555.

Doucet S., Scoates J. S., Weis D. and Giret A. (2005) Constrainingthe components of the Kerguelen mantle plume: A Hf-Pb-Sr-Nd isotopic study of picrites and high-MgO basalts from theKerguelen Archipelago. Geochem. Geophys. Geosyst. 6, n/a–n/a.

Durr H. H., Meybeck M. and Durr S. H. (2005) Lithologiccomposition of the Earth’s continental surfaces derived from anew digital map emphasizing riverine material transfer. GlobalBiogeochem. Cycles 19, 1–22.

Flament N., Coltice N. and Rey P. F. (2013) The evolution of the87Sr/86Sr of marine carbonates does not constrain continentalgrowth Elsevier B.V.. Precambrian Res. 229, 177–188.

Foley B. J. (2018) The dependence of planetary tectonics on mantlethermal state: applications to early Earth evolution. Philosoph.Trans. Roy. Soc. A: Math. Phys. Eng. Sci. 376, 20170409–20170430.

Foley S. F., Tiepolo M. and Vannucci R. (2002) Growth of earlycontinental crust controlled by melting of amphibolite insubduction zones. Nature 417, 837–840.

Gale A., Dalton C. A., Langmuir C. H., Su Y. and Schilling J.-G.(2013) The mean composition of ocean ridge basalts. Geochem.

Geophys. Geosyst. 14, 489–518.Garcia D., Coelho J. and Perrin M. (1991) of sorting within shale

and sandstone series (northern Portugal). Eur. J. Mineral. 3,401–414.

Garcia D., Fonteilles M. and Moutte J. (1994) Sedimentaryfractionations between Al, Ti, and Zr and the genesis ofstrongly peraluminous granites. J. Geol. 102, 411–422.

Garcon M., Chauvel C. and France-Lanord C. (2013a) Continentalsedimentary processes decouple Nd and Hf isotopes. Geochim.

Cosmochim. Acta 121, 177–195.Garcon M., Chauvel C., France-Lanord C., Huyghe P. and Lave J.

(2013b) Continental sedimentary processes decouple Nd and Hfisotopes Elsevier Ltd. Geochim. Cosmochim. Acta 121, 177–195.

Garcon M., Chauvel C., France-Lanord C., Limonta M. andGarzanti E. (2014) Which minerals control the Nd–Hf–Sr–Pbisotopic compositions of river sediments? Elsevier BV.. Chem.

Geol. 364, 42–55.Garzanti E., Ando S., France-Lanord C., Censi P., Vignola Pietr,

Galy V. and Lupker M. (2011) Mineralogical and chemicalvariability of fluvial sediments 2. Suspended-load silt (Ganga–Brahmaputra, Bangladesh) Elsevier B.V.. Earth Planet. Sci.

Lett. 302, 107–120.Gaschnig R. M., Rudnick R. L., McDonough W. F., Kaufman A.

J., Valley J. W., Hu Z., Gao S. and Beck M. L. (2016)Compositional evolution of the upper continental crust throughtime, as constrained by ancient glacial diamictites Elsevier Ltd.Geochim. Cosmochim. Acta 186, 316–343.

GeoRoc., <http://georoc.mpch-mainz.gwdg.de/georoc/>, accessedNovember 2018.

Greber N. D., Dauphas N., Bekker A., Ptacek M. P., Bindeman I.N. and Hofmann A. (2017a) Titanium isotopic evidence forfelsic crust and plate tectonics 3.5 billion years ago. Science 357,1271–1274.

Greber N. D., Dauphas N., Puchtel I. S., Hofmann B. A. andArndt N. (2017b) Titanium stable isotopic variations inchondrites, achondrites and lunar rocks Elsevier Ltd. Geochim.

Cosmochim. Acta 213, 534–552.Grove T. L., Till C. B. and Krawczynski M. J. (2012) The role zone

magmatism. Annu. Rev. Earth Planet. Sci. 40, 413–439.


http://refhub.elsevier.com/S0016-7037(19)30229-7/h0005















































































































http://georoc.mpch-mainz.gwdg.de/georoc/












Hamelin C., Seitz H.-M., Barrat J.-A., Dosso L., Maury R. C. andChaussidon M. (2009) A low d7Li lower crustal component:Evidence from an alkalic intraplate volcanic series (Chaıne desPuys, French Massif Central). Chem. Geol. 266, 205–217.

Hayashi K. I., Fujisawa H., Holland H. D. and Ohmoto H. (1997)Geochemistry of similar to �1.9 Ga sedimentary rocks fromnortheastern Labrador, Canada. Geochim. Cosmochim. Acta 61,4115–4137.

Hartmann J. and Moosdorf N. (2012) The new global lithologicalmap database GLiM: A representation of rock properties at theEarth surface. Geochem. Geophys. Geosyst. 13, 119–137.

Hawkesworth C. J. and Brown M. (2018) Earth dynamics and thedevelopment of plate tectonics. Philosoph. Trans. Roy. Soc. A:Math. Phys. Eng. Sci. 376, 20180228–20180235.

Hawkesworth C., Cawood P., Kemp T., Storey C. and Dhuime B.(2009) A matter of preservation. Science 323, 49–50.

Hildreth W. and Moorbath S. (1988) Crustal contributions to arcmagmatism in the Andes of central Chile. Contrib. Miner.

Petrol. 98, 455–489.Hooper P. R. (2000) Chemical discrimination of Columbia River

basalt flows. Geochem. Geophys. Geosyst. 1. https://doi.org/10.1029/2000GC000040.

Hopkins M. D., Harrison T. M. and Manning C. E. (2010)Constraints on Hadean geodynamics from mineral inclusions in>4Ga zircons Elsevier B.V.. Earth Planet. Sci. Lett. 298, 367–376.

Jagoutz O. and Kelemen P. B. (2015) Role of Arc Processes in theformation of continental crust. Annu. Rev. Earth Planet. Sci. 43,363–404.

Jagoutz O. E. (2010) Construction of the granitoid crust of anisland arc. Part II: a quantitative petrogenetic model. Contrib.Miner. Petrol. 160, 359–381.

Jagoutz O. E., Burg J. P., Hussain S., Dawood H., Pettke T., IizukaT. and Maruyama S. (2009) Construction of the granitoid crustof an island arc part I: geochronological and geochemicalconstraints from the plutonic Kohistan (NW Pakistan). Con-trib. Miner. Petrol. 158, 739–755.

Kelemen P. B., Hanghøj K. and Greene A. R. (2014) One view ofthe geochemistry of subduction-related magmatic arcs, with anemphasis on primitive andesite and lower crust. Treatise

Geochem., 749–806.Kelemen P. B., Johnson K., Kinzler R. J. and Nature A. I. (1990)

High-field-strength element depletions in arc basalts due tomantle–magma interaction. Nature 345, 521–524.

Keller B. and Schoene B. (2018) Plate tectonics and continentalbasaltic geochemistry throughout Earth history Elsevier B.V..Earth Planet. Sci. Lett. 481, 290–304.

Keller C. B. and Schoene B. (2012) Statistical geochemistry revealsdisruption in secular lithospheric evolution about 2.5 Gyr agoNature Publishing Group. Nature 485, 490–493.

Keller C. B., Boehnke P. and Schoene B. (2017) Temporal variationin relative zircon abundance throughout Earth history.Geochem. Perspect. Lett., 179–189.

Korenaga J. (2018) Crustal evolution and mantle dynamicsthrough Earth history. Philosoph. Trans. Roy. Soc. A: Math.

Phys. Eng. Sci. 376, 20170408–20170435.Large R. R., Mukherjee I., Zhukova I., Corkrey R., Stepanov A.

and Danyushevsky L. V. (2018) Role of upper-most crustalcomposition in the evolution of the Precambrian ocean–atmosphere system Elsevier B.V.. Earth Planet.Sci. Lett. 487,44–53.

Le Bas M. J. and Streckeisen A. L. (1991) The IUGS systematics ofigneous rocks. J. Geol. Soc. 148, 825–833.

Lee C.-T. A., Yeung L. Y., McKenzie N. R., Yokoyama Y., OzakiK. and Lenardic A. (2016) Two-step rise of atmospheric oxygen

linked to the growth of continents Nature Publishing Group.Nature Geosci. 9, 1–10.

Lenardic A. (2018) The diversity of tectonic modes and thoughtsabout transitions between them. Philosoph. Trans. Roy. Soc. A:Math. Phys. Eng. Sci. 376, 20170416–20170422.

Li X., Li J., Bader T., Mo X., Scheltens M., Chen Z., Xu J., Yu X.and Huang X. (2015) Evidence for crustal contamination inintra-continental OIB-like basalts from West Qinling, centralChina: A Re–Os perspective. J. Asian Earth Sci. 98, 436–445.

Lupker M., France-Lanord C., Lave J., Bouchez J., Galy V.,Metivier F., Gaillardet J., Lartiges B. and Mugnier J.-L. (2011)A Rouse-based method to integrate the chemical compositionof river sediments: Application to the Ganga basin. Geophys.Res. Lett. 116, 488–524.

Lyons T. W., Reinhard C. T. and Planavsky N. J. (2014) The riseof oxygen in Earth’s early ocean and atmosphere. Nature 506,307–315.

McLennan S. M., Hemming S., McDaniel D. K. and Hanson G. N.(1993) Geochemical approaches to sedimentation, provenance,and tectonics. Geol. Soc. Am. Spec. Pap. 284, 21–40.

Millet M.-A. and Dauphas N. (2014) Ultra-precise titanium stableisotope measurements by double-spike high resolution MC-ICP-MS. J. Anal. At. Spectrom. 29, 1444.

Millet M.-A., Dauphas N., Greber N. D., Burton K. W., Dale C.W., Debret B., Macpherson C. G., Nowell G. M. and WilliamsH. M. (2016) Titanium stable isotope investigation of magmaticprocesses on the Earth and Moon Elsevier B.V.. Earth Planet.

Sci. Lett. 449, 197–205.Moosdorf N., Hartmann J. and Durr H. H. (2010) Lithological

composition of the North American continent and implicationsof lithological map resolution for dissolved silica flux modeling.Geochem. Geophys. Geosyst. 11, n/a–n/a.

Moyen J.-F. and Laurent O. (2018) Archaean tectonic systems: Aview from igneous rocks. Lithos 302–303, 99–125.

Moyen J.-F. and Martin H. (2012) Forty years of TTG researchElsevier B.V.. Lithos 148, 312–336.

Munker C., Worner G., Yogodzinski G. and Churikova T. (2004)Behaviour of high field strength elements in subduction zones:constraints from Kamchatka-Aleutian arc lavas. Earth Planet.

Sci. Lett. 224, 275–293.Muntener O. and Ulmer P. (2018) Arc crust formation and

differentiation constrained by experimental petrology. Am. J.

Sci. 318, 64–89.PetDB., <https://www.earthchem.org/petdb>, accessed November

2018.Polat A., Wang L. and Appel P. W. U. (2015) A review of

structural patterns and melting processes in the Archean cratonof West Greenland: Evidence for crustal growth at convergentplate margins as opposed to non-uniformitarian models Else-vier B.V.. Tectonophysics 662, 67–94.

Powell D. M. (1998) Patterns and processes of sediment sorting ingravel-bed rivers. Prog. Phys. Geogr. 22, 1–32.

Prytulak J. and Elliott T. (2007) TiO2 enrichment in ocean islandbasalts. Earth Planet. Sci. Lett. 263, 388–403.

Puchtel I. S., Blichert-Toft J., Touboul M., Walker R. J., Byerly G.R., Nisbet E. G. and Anhaeusser C. R. (2013) Insights intoearly Earth from Barberton komatiites: Evidence from litho-phile isotope and trace element systematics. Geochim. Cos-

mochim. Acta 108, 63–90.Reimink J. R., Chacko T., Carlson R. W., Shirey S. B., Liu J.,

Stern R. A., Bauer A. M., Pearson D. G. and Heaman L. M.(2018) Petrogenesis and tectonics of the Acasta Gneiss Complexderived from integrated petrology and 142Nd and 182W extinctnuclide-geochemistry Elsevier B.V.. Earth Planet.Sci. Lett. 494,12–22.






















https://doi.org/10.1029/2000GC000040

https://doi.org/10.1029/2000GC000040























































































https://www.earthchem.org/petdb























Rey P. F. and Coltice N. (2008) Neoarchean lithospheric strength-ening and the coupling of Earth’s geochemical reservoirs.Geology 36, 635–638.

Roberts N. M. W. and Spencer C. J. (2015) The zircon archive ofcontinent formation through time London, Special Publica-tions. Geol. Soc. 389, 197–225.

Rudnick R. L. and Gao S. X. (2003) Composition of thecontinental crust. Treatise Geochem., 659.

Sizova E., Gerya T., Stuwe K. and Brown M. (2015) Generation offelsic crust in the Archean: A geodynamic modeling perspectiveElsevier B.V.. Precambrian Res. 271, 198–224.

Smart K. A., Tappe S., Stern R. A., Webb S. J. and Ashwal L. D.(2016) Early Archaean tectonics and mantle redox recorded inWitwatersrand diamonds. Nat. Geosci. 9, 255–259.

Smit M. A. and Mezger K. (2017) Earth’s early O2 cycle suppressedby primitive continents. Nat. Geosci. 10, 788–792.

Tang M., Chen K. and Rudnick R. L. (2016) Archean upper crusttransition from mafic to felsic marks the onset of platetectonics. Science 351, 372–375.

Taylor S. R. and McLennan S. M. (1985) The Continental Crust:

Its Composition and Evolution. Blackwell Scientific, Oxford.Ulmer P. (2001) Partial melting in the mantle wedge - the role of

genesis of mantle-derived ‘‘arc-related” magmas. Phys. EarthPlanet. Inter. 127, 215–232.

Van Kranendonk M. J., Smithies R. H., Griffin W. L., Huston D.L., Hickman A. H., Champion D. C., Anhaeusser C. R. andPirajno F. (2015) Making it thick: a volcanic plateau origin ofPalaeoarchean continental lithosphere of the Pilbara andKaapvaal cratons London, Special Publications. Geol. Soc.

389, 83–111.Vanderkluysen L., Mahoney J. J., Hooper P. R., Sheth H. C. and

Ray R. (2011) The feeder system of the Deccan traps (India):Insights from dike geochemistry. J. Petrol. 52, 315–343.

Voice P. J., Kowalewski M. and Eriksson K. A. (2011) Quantifyingthe timing and rate of crustal evolution: global compilation ofradiometrically dated detrital zircon grains. J. Geol. 119, 109–126.

Wedepohl H. K. (1995) The composition of the continental crust.Geochim. Cosmochim. Acta 59, 1217–1232.

Woodhead J. D., Eggins S. M. and Johnson R. W. (1998) Magmagenesis in the New Britain island arc: Further insights intomelting and mass transfer processes. J. Petrol. 39, 1641–1668.

Worsley T. R., Nance D. and Moody J. B. (1984) (1984). Globaltectonics and eustasy for the past 2 billion years. Mar. Geol. 58,373–400.

Associate editor: Aubrey Zerkle
















































The chemistry of fine-grained terrigenous sediments ...The chemistry of ﬁne-grained terrigenous...

Documents

Transcript of The chemistry of fine-grained terrigenous sediments ...The chemistry of ﬁne-grained terrigenous...