A High Continental Weathering Flux Into Paleoarchean Seawater Revealed

Earth and Planetary Science Letters 454 (2016) 28–35

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

www.elsevier.com/locate/epsl

A high continental weathering ﬂux into Paleoarchean seawater revealed by strontium isotope analysis of 3.26 Ga barite

a,b, c a,b b,d Aaron M. Satkoski ∗, Donald R. Lowe , Brian L. Beard , Max L. Coleman , Clark M. Johnson a,b a University of Wisconsin–Madison, Department of Geoscience, 1215 West Dayton Street, Madison, WI 53706, United States b NASA Astrobiology Institute, United States c Stanford University, Department of Geological and Environmental Sciences, 118 Braun Hall, Stanford, CA 94305, United States d Planetary Surface Instruments Group, NASA Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, United States a r t i c l e i n f o a b s t r a c t

Article history: Controls on Archean seawater chemistry remain controversial. Many studies have suggested that it Received 11 March 2016 was largely controlled by oceanic hydrothermal fluid circulation. Recent work, however, from clastic Received in revised form 11 August 2016 sequences, Hf–O isotope data from detrital zircons, and models for the Rb/Sr evolution of the continental Accepted 26 August 2016 crust suggest that intense continental weathering and low-temperature surface alteration were more Available online xxxx important than previously thought during the early Archean. This is consistent with biogeochemical Editor: M. Bickle studies that suggest the Archean had a diverse microbial ecology, which would, in part, need to be Keywords: sustained by nutrients (e.g., phosphorus) that were derived from continental weathering. To further Archean seawater quantify continental weathering during the early Archean, we analyzed 3.26 Ga barite from the Fig Tree strontium isotopes Group, South Africa for strontium, oxygen, and sulfur isotope compositions. We propose that the seawater barite component of the barite is characterized by 87Sr/86Sr ratios >0.701, which is significantly more radiogenic than contemporaneous mantle ( 0.7007–0.7008). The radiogenic nature of seawater at this time suggests ∼ that the continental weathering flux at 3.26 Ga had a large impact on ocean chemistry 400 million years earlier than previously suggested.  2016 Elsevier B.V. All rights reserved.

1. Introduction that continental weathering during the early Archean was insignif- icant, where the majority of continental crust was submerged and Understanding the temporal changes in the volume and com- relatively mafic in composition, which in turn would suggest that position of continental crust over Earth history bears on models seawater chemistry was largely controlled by oceanic hydrother- for crust–mantle evolution, as well as changes in the surface envi- mal fluid circulation (Shields and Veizer, 2002; Shields, 2007; ronments of the Earth, including the biosphere. For example, CO2 Flament et al., 2013). This view, however, must be re-evaluated in the modern atmosphere is consumed during chemical weath- based on recent studies that address continental evolution and ering (Beaulieu et al., 2012). Carbon dioxide was a very large alteration in the early Earth. Lithium isotope data from the Jack component of the Hadean–Archean atmosphere (Kasting, 2014); Hills zircons suggest that continental-like crust was weathering therefore, continental weathering likely had a significant impact on as early as 4300 Ma (Ushikubo et al., 2008). Hafnium and O iso- the climate during this time. In addition, studies of biogeochemi- tope studies of detrital zircons document a strong increase in re- cal cycles recorded in early Archean sedimentary rocks suggest an cycling of evolved continental crust at 3.2 Ga, commensurate ∼18 early diverse microbial ecology that may have required extensive with a decrease in εHf and increase in δ Ovalues, suggesting continentally-derived nutrients such as phosphorus early in Earth that low-temperature alteration of evolved continental crust was history (Blake et al., 2010); such a proposal would require the pres- more widespread throughout the Archean than previously thought ence of emergent, evolved crust. In contrast, it has been argued (Dhuime et al., 2012). Such a proposal is consistent with models that suggest extensive weathering conditions existed throughout the Archean (Hessler and Lowe, 2006 and references therein). This recent work, therefore, requires a re-evaluation of the tradi- * Corresponding author at: University of Wisconsin–Madison, Department of Geo- science, 1215 West Dayton Street, Madison, WI 53706, United States. tional view that emergent continental crust was minor in the early E-mail address: [email protected] (A.M. Satkoski). Archean. http://dx.doi.org/10.1016/j.epsl.2016.08.032 0012-821X/ 2016 Elsevier B.V. All rights reserved. A.M. Satkoski et al. / Earth and Planetary Science Letters 454 (2016) 28–35 29

A common approach for assessing the continental flux to seawater has been the use of 87Sr/86Sr ratios on marine carbonate (e.g., Shields and Veizer, 2002), which depends on the balance between the two primary sources of Sr to seawater: (1) oceanic hydrothermal fluids and (2) continental weathering. Although Sr isotope seawater curves have been defined by very large databases for the Phanerozoic, the seawater Sr isotope curve for early- to mid-Archean rocks has been problematic because pristine marine carbonate is rare. As shown by Shields and Veizer (2002) and Prokoph et al. (2008), very few samples define the Archean por- tion of the seawater curve, except between 2800–2700 Ma. Al- though the density of data for carbonates of this age is high, the range in 87Sr/86Sr ratios is very large (0.70114–>0.708). Restriction to the least radiogenic compositions (87Sr/86Sr 0.7011) is one ∼ approach to placing estimates on seawater compositions, based on the assumption that the presence of detrital components or later alteration will increase 87Sr/86Sr ratios (Shields and Veizer, 2002). The least radiogenic Sr isotope compositions for carbonates of 2800–2700 Ma age, however, all have relatively low δ18O (VSMOW) values that range from 9.9–17.6 (Prokoph et al., 2008), suggesting that hydrothermal fluids may have altered these samples; this, in fact, suggests that the common approach of using Fig. 1. (A) Map showing the location of the southwestern half of the Barberton the least radiogenic carbonate to infer seawater, as is generally greenstone belt. (B) Geologic map of the southwestern half of the Barberton greenstone belt. Number 1 on the map shows the location of the Barite Valley section and done for the Phanerozoic, may be flawed, and under-estimate the number 2 shows the location of the Conglomerate Quarry section. Detailed sample 87 86 Sr/ Sr ratios of Archean seawater. location information can be found in Supplementary Table S1. Map is modified from With little pristine carbonate to define it, there is much un- Lowe (2013). certainty in the Sr isotope composition of Paleoarchean seawater. Given recent modeling of large databases for igneous rocks which the shallow-water parts of the Fig Tree Group sequence. Barite oc- suggests that the Rb/Sr ratios of the continental crust began a con- curs in two morphologies: granular and bladed, which are always tinued rise from 3.2 Ga to the end of the Precambrian (Dhuime found in close association (Fig. 2). The granular type is associ- ∼ et al., 2015), it is important to re-evaluate the Sr isotope seawa- ated with detrital grains of pyrite, Cr-spinel, quartz, zircon, and ter curve for this time interval. If Rb/Sr ratios of the continental muscovite, and has been interpreted as a locally reworked pri- crust started to rise at 3.2 Ga, and this crust was emergent, mary precipitate (Heinrichs and Reimer, 1977). The granular barite ∼ an inevitable outcome would be an increase in the 87Sr/86Sr ra- does not likely represent erosion, transport, and deposition from tio of the continents. Here we re-visit the early Archean seawater older barite deposits of the Onverwacht Group (Reimer, 1980)be- Sr isotope curve, using barite, a relatively insoluble mineral that is cause if erosion had stripped the 4–5 km of intervening strata the resistant to isotopic exchange by later alteration as compared to deposited sediments would be much thicker and more varied in carbonates (see review of Bao, 2015). Although not common in the composition, and the barite would be a negligible component. The Archean geologic record, marine barite has the potential to cap- bladed type can be observed to drag chert bedding upward or ture the 87Sr/86Sr ratio of seawater (Griffith and Paytan, 2012 and bedding can sag around a blade, and both textures suggest that references therein), and offers an alternative to marine carbonates the barite is syn-diagenetic and grew early in the sedimentary as a record for seawater chemistry. Our target is marine stratiform sequence (Heinrichs and Reimer, 1977). We interpret the bladed barite deposits of the 3.2 Ga Fig Tree Group, South Africa, and we morphology to reflect a separate, primary phase. No evidence ex- combine Sr, O, and S isotopes to distinguish between ambient sea- ists that the barite blades formed as a replacement after gypsum water and hydrothermal Sr sources, with the goal of providing, for or that it was re-crystallized by low-grade metamorphism (Bao et the first time, a robust datum on the seawater Sr isotope curve in al., 2007). the Paleoarchean. 3. Methods 2. Geologic background 3.1. Sampling The Barberton greenstone belt (BGB) is comprised, from old- est to youngest, of three main lithostratigraphic units: the On- Samples are from two different stratigraphic positions within verwacht Group (primarily volcanic), the volcaniclastic and silici- the Fig Tree Group (Fig. 3). The stratigraphically lowest samples are clastic Fig Tree Group, and the siliciclastic Moodies Group (Lowe, typically within 1–5 m of the contact with the cherts at the top of 2013; Fig. 1). In the Barite Valley area in the central greenstone the Mendon Formation in the Onverwacht Group (Fig. 3). These belt, dacitic tuffs from the Mapepe Formation, principal unit of the cherts, 40–100 m thick, overlie a thick sequence of Onverwacht Fig Tree Group in the middle and southern parts of the greenstone komatiites. Samples from here include both the bladed and gran- belt, have yielded U–Pb zircon ages between about 3260 Ma near ular morphology. A second suite of samples was analyzed from the base of the unit to about 3225 Ma from near the top (Kröner the middle Mapepe Formation, where barite is interbedded with et al., 1991). dacitic and siliciclastic sediments (Fig. 3). Barites sampled from the The Fig Tree Group represents a variety of depositional environ- middle Mapepe Formation all have a bladed morphology. ments that range from shallow to deep subaqueous, alluvial and The barite samples were cut and polished into flat slabs, from fan-delta. Barite layers are widely developed in the lower part of which specific areas could be sampled with a high spatial resolu- the Fig Tree Group, where they form discontinuous lenses primar- tion, including granular-rich and blade-rich morphologies (Fig. 2). ily in the Barite Valley and Conglomerate Quarry areas (Heinrichs Detailed sampling was done using a New Wave Micromill in raster and Reimer, 1977; Fig. 1). These barite layers are contained within mode, with special care taken to avoid any areas that appeared 30 A.M. Satkoski et al. / Earth and Planetary Science Letters 454 (2016) 28–35

Fig. 2. (A) Photograph showing a representative hand sample of barite analyzed as part of this study. Highlighted are the two primary barite morphologies, bladed and granular. The granular barite is recognized by its green color, due to the presence of weathered Cr-spinel. The dashed areas show where samples were micro drilled as part of this study. (B) Representative plain light photomicrograph showing the relationship between the bladed and granular barite. (C) Representative secondary electron image showing the bladed and granular barite. Cr-spinel is commonly found associated with the granular barite. Chert fills the space in between granular barite and the detrital minerals. No significant inclusions are observed in barite blades or granules. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) to have undergone post-formation alteration. Milled barite powder <0.000035, indicating that the measured 87Sr/86Sr ratio may be (0.001–0.009 g) was split into different aliquots for Sr, O, and S taken as the initial 87Sr/86Sr ratio. isotope analyses. 3.3. Oxygen and sulfur isotope analysis 3.2. Strontium isotope analysis Oxygen isotope analyses were performed on approximately The barite split for the Sr isotope analysis was spiked with an 0.15 mg BaSO4 weighed into silver capsules, reduced by glassy enriched 87Rb–84Sr tracer and dissolved using 99.999% trace metal carbon in a Thermo Finnigan High Temperature Conversion Ele- grade Na2CO3 from Sigma-Aldrich following the method of Breit et mental Analyzer (TC/EA) at 1450 ◦C to produce CO passed through al. (1985). A Sr blank of 35 ppb was measured for the Na2CO3, and a Thermo Conflo III system and measured via CF-IRMS on a MAT with sample concentrations >1000 ppm, the Sr added from the 253 mass spectrometer (Waltham MA, USA), at the Jet Propulsion 87 86 18 Na2CO3 does not affect the measured Sr/ Sr ratio. Strontium Laboratory, California. δ Ovalues are reported in δ notation in per from the barite samples was isolated using Sr-Spec® resin and mil ( ) relative to Vienna Standard Mean Ocean Water (VSMOW). 18 Rb was isolated using BioRad AX-50 ion-exchange resin. Strontium Reproducibilityh of δ Omeasurements was checked each run us- was analyzed using a VG Sector 54 thermal ionization mass spec- ing laboratory standards of BaSO4 and national and international trometer in multidynamic mode at the University of Wisconsin– standards NBS 127, and IAEA SO-6 and is better than 0.5 (2σ ). Madison. Samples were loaded on single Ta filaments with H3PO4. Accuracy of internal BaSO4 standards were checked using IAEAh 601, 11 Total Sr ion intensities were 3.85 10− A, and the average NBS 127 and IAEA SO-6, and is better than 0.5 2σ . 87 86 ∼ × Sr/ Sr ratio is based on 120 ratios. The Sr isotope standard Sulfur isotope compositions were measuredh by taking approx. NIST SRM-987 analyzed during the time period (n 83) that barite 0.5 mg BaSO weighed into tin capsules, combusted with oxygen in = 4 analyses were collected was 0.710271 18 (2-SD, external). a Costech Elemental Combustion System (Valencia CA, USA) to pro- ± Rubidium measurements were made in static mode or done as duce SO2 passed through a Thermo Conflo III system and measured a peak hop on a single Faraday collector and the average 87Rb/85Rb by CF-IRMS using a Thermo MAT 253 mass spectrometer (Waltham ratio is based on 50 measured ratios. The measured 87Rb/85Rb ratio MA, USA) at the Jet Propulsion Laboratory, California. Sulfur isotope of NIST SRM-984 over the course of when barite was analyzed was compositions are reported in δ notation in per mil ( ) relative 0.3852 0.0024 (n 17; 2-SD). Measured Sr isotope ratios were to Canyon Diablo Troilite (CDT). Reproducibility of δ34Smeasure- ± = h corrected to initial ratios (3.26 Ga) based on the depositional age of ments was checked each run using internal standards of BaSO4, 87 86 the Fig Tree Group. Typical Rb/ Sr ratios for the samples are less Ag2 S, and elemental S, and is better than 0.3 (2σ ). Accuracy of than 0.00075, making the correction to the initial ratio at 3.26 Ga internal standards were checked using nationalh and international A.M. Satkoski et al. / Earth and Planetary Science Letters 454 (2016) 28–35 31

Fig. 3. A representative stratigraphic section for the Mapepe Formation modiﬁed from Lowe and Nocita (1999). Shown are histograms of the Sr, O and S isotope data separated out by their stratigraphic position, either in the lower or middle Mapepe formation. standards NBS 127, IAEA S-2, and IAEA S-1, whose reproducibility weighted average δ18Ovalue for these samples is 8.3 2.0 (2σ ), ± is better than 0.2 (2σ ). which is similar to the bladed barite from lower in the section. The 18 34 The range of replicateh δ Oand δ Sanalyses of samples was weighted average δ34Svalue for these samples is 4.8 0.4 (2σ ), ± noticeably greater than that of known homogeneous standards (Ta- which is similar to both the bladed and granular barite from lower ble S-1). The purity of samples in some cases showed that they in the section. The range in δ34Svalues of 2.6 to 7.5 for all sam- were not pure barite and probably contained minor chert, which ples studied here is similar to the values reported by Bao et al. is found with the barite in the samples (Fig. 2). It is unlikely (2007) and Roerdink et al. (2012) for Fig Tree Group barite. Sim- that chert contributed signiﬁcantly to the measured oxygen iso- ilarly, the range in δ18Ovalues of 6.3 to 12.9 for all samples is tope compositions because only minimal chert would have been comparable to that reported for barite from the Fig Tree Group by converted to CO during sample processing, and the amount of Bao et al. (2007), although the results form the current study pro- chert contamination was quite low. Therefore, the replicate mea- duced a larger range than that measured by Bao et al. (2007) of surement range most likely represents sample heterogeneity rather (9.3–12.9). that analytical uncertainty. 5. Discussion 4. Results

The initial 87Sr/86Sr ratios for barite from the lower Mapepe Most Archean barite is thought to have formed by mixing of Formation typically range from 0.7011–0.7016, with only one Ba-rich (hydrothermal) fluids with sulfate-bearing seawater (Jewell, sample that has a ratio of 0.7024 (Fig. 3). This range overlaps 2000; Huston and Logan, 2004). This is consistent with previ- 33 with that determined by Perry et al. (1971) and Gutzmer et ous work that has shown that Fig Tree barite samples have $ S al. (2006) for barite from the Fig Tree Group (0.70088–0.70172 anomalies, suggesting that seawater was the primary source for and 0.7009–0.7017, respectively), which Perry et al. (1971) inter- sulfate (Bao et al., 2007; Roerdink et al., 2012), which initially preted to record an igneous source. The weighted average initial likely formed through atmospheric photolysis reactions. The Ba- 87Sr/86Sr ratio for the bladed barite is 0.70135 24 (2σ ), whereas rich fluids are interpreted to have been vented through crustal ± the weighted average initial 87Sr/86Sr ratio of the granular barite fractures generated by one or more large meteorite impacts (Lowe, is 0.70139 9(2σ ). The weighted average δ18Ovalue for the 2013). The crustal fractures are abundant in the Barite Valley, ± bladed barite is 8.7 0.7(2σ ). The granular barite, however, has and crosscut the komatiites of the top of the Onverwacht Group ± a higher weighted average δ18Ovalue of 10.3 1.0 (2σ ). The (3.33–3.26 Ga Mendon Formation). These observations suggest that ± weighted average δ34Svalue for the bladed and granular barites the barite likely contains at least two components, one derived are 4.5 1.0 (2σ ) and 3.9 1.0 (2σ ), respectively. from seawater and one that reflects hydrothermal fluid–rock in- ± ± The initial 87Sr/86Sr ratios for samples from the middle Mapepe teraction with basement rocks. The hydrothermal (Ba-rich) compo- Formation range from 0.7014–0.7042 and have a weighted aver- nent probably records high-temperature fluid–rock interaction at age of 0.7022 0.0011 (2σ ), which is significantly more radio- depth in the chert dike system, followed by cooling during barite ± genic than the samples from the lower Mapepe Formation. The formation at the seawater–rock interface. 32 A.M. Satkoski et al. / Earth and Planetary Science Letters 454 (2016) 28–35

the barite to assess hydrothermal (low δ18O) versus seawater (high δ18O) components. The effects of hydrothermal fluid–rock interaction are clearly seen in samples from the middle Mapepe Formation, and a highly metamorphosed sample, that have very high 87Sr/86Sr ratios (Fig. 4B). The metamorphosed barite sample (sample 540-1), which is not located near the primary sampling areas shown in Fig. 1, re- tains no primary textures, suggesting it may have been altered during recrystallization. The middle Mapepe Formation contains felsic volcanic and clastic rocks, whose source regions had εNd values that range from 1.6 to 2.5 at 3.2 Ga (Clemens et al., 2006), − − and this, in addition to detrital zircon populations that have pre- depositional ages of 3530 to 3220 Ma (Zeh et al., 2013), suggests that the middle Mapepe felsic volcanic and clastic rocks likely had elevated 87Sr/86Sr ratios at the time of barite formation. Although δ18Ovalues have not been measured for the middle Mapepe clastic rocks, the generally high CAI values for Fig Tree Group shales ( 63–82; Eriksson and Soegaard, 1985) suggests that they should ∼ have elevated δ18Ovalues. Thus, we propose the barite samples from the middle Mapepe Formation formed, in part, from Ba-rich fluids that heavily interacted with the underlying felsic volcanic and clastic rocks. Therefore, these samples have been heavily in- fluenced by crustal O and Sr that was obtained during precursor hydrothermal circulation, and place no constraints on seawater. In contrast, underlying samples of the lower Mapepe Formation are the mafic rocks of the Mendon Formation (Fig. 3), distinctly different than the middle Mapepe Formation. Barium-rich fluids that permeated through the top of the mafic Mendon Formation via the cross-cutting chert dikes would have had near-mantle δ18Ovalues, due to high-temperature exchange with mafic (low-δ18O) lithologies. Furthermore, the 87Sr/86Sr ratios of this fluid would approach that of underlying komatiite, 0.7007, as calculated by extrapolat- ∼ ing the measurements of underlying komatiite from Jahn and Shih (1974) using a depleted-mantle 87Rb/86Sr ratio of 0.07 (McCulloch, 1994). The isotope ratio from the underlying komatiite is very sim- Fig. 4. (A) Cross plot of δ18Oversus δ34Sfor bladed and granular barite. A lin- ilar to estimates for the mantle at this time (McCulloch, 1994; ear trend (not shown) through these data produces an R2 of 0.05, which indicates Kamber, 2010). The low-87Sr/86Sr ratios and low-δ18Ovalues of 18 87 86 no statistically meaningful trend exists. (B) Cross plot of δ Oversus Sr/ Sr the lower Mapepe sample suite may therefore not reflect those for bladed and granular barite. The light grey box represents the weighted mean of seawater, but instead a Ba-rich fluid that inherited the Sr iso- (95% conf.) of δ18O (10 1 ) and 87Sr/86Sr (0.70139 9) for the granular barite. ± ± The range for mantle δ18Oish 5.3 0.6 (Dhuime et al., 2012)and mantle Sr is tope composition of the underlying komatiites. This conclusion is ± 18 0.7007–0.7008 (Kamber and Webb, 2001), which is shown with the dark grey box. most strongly supported by the very low δ O values of 7, which h ∼ The line that we draw is the line that exists between the isotope composition of the essentially require a hydrothermal component. Such a conclusion mantle at that time and (based on the granular barite) the isotope composition of demonstrates that in cases where a hydrothermal component was what we think the best estimate for seawater is at the time. Most of the data from the lower Mapepe Formation fall along this line. involved, it may not be safe to assume that the least radiogenic Sr isotope composition reflects the best estimate for seawater Sr. This point was made by McCulloch (1994) in a study of 3.45 Ga barite 5.1. Distinguishing seawater and hydrothermal components from the Pilbara craton, Australia, which was interpreted to reflect the mantle and not ambient seawater. The lack of correlation between δ34Sand δ18O values (Fig. 4A) The granular barite samples of the lower Mapepe Formation 18 is interpreted to reflect the more sluggish isotopic kinetics of S that have the highest δ Ovalues should reflect Sr isotope compositions that are closest to seawater. This follows the petrographic in sulfate relative to O (Chiba and Sakai, 1985), suggesting that S evidence that granular barite is most likely to record ambient sea- would tend to maintain its initial isotope composition. According water conditions. This in turn suggests that seawater sulfate (not to modeling by Philippot et al. (2007), photolysis-generated sul- ocean water) at the time had a δ18Ovalue of 10 1 . In con- fate could have δ34Svalues of 3 to 4, which overlaps the lower ∼ ± ∼ trast, the lower δ18Ovalues, as recorded in the range observed end of the range measured in the barite studied here. Seawater h for the bladed barite from the lower Mapepe sample suite, is in- sulfate in the Early Archean is thought to have had δ34Svaluesbe- terpreted to reflect high-temperature exchange with the mafic ig- tween 3 and 7 (Canfield and Farquhar, 2009), which matches the neous lithologies that underlie the base of the Fig Tree Group, as ∼ 18 range observed in the barites of this study. The exact δ Ovalue noted above. The 87Sr/86Sr ratio of the granular barite from the for sulfate generated by photolysis is unknown (Bao et al., 2007), lower Mapepe Formation has a weighted average of 0.70139 9 ± although the oxygen isotope composition of sulfate is temperature (2σ ), indicating that seawater, at a minimum, was significantly dependent, and hydrothermal contributions at temperatures above more radiogenic than the mantle at this time ( 0.7007; Shields 18 ∼ 150 ◦C would produce δ Ovalues that would be significantly and Veizer, 2002). Although the Sr isotope contrast we propose ∼ lower than those of seawater sulfate (Chiba and Sakai, 1985), de- between seawater and mantle at 3.2 Ga may seem small, it is rela- pendent upon the temperatures of fluid–rock interaction and the tively large considering the limited range in 87Sr/86Sr ratios in the lithologies involved. We therefore focus on use of δ18Ovalues of Earth at this time. A.M. Satkoski et al. / Earth and Planetary Science Letters 454 (2016) 28–35 33

derived hydrothermal fluids or mafic igneous rocks, such as the underlying komatiite. These results, therefore, provide a first or- der observation that a significantly larger flux of evolved continental material was delivered to the ocean at this time, relative to previous assessments of Sr isotope compositions in seawater (e.g., Shields and Veizer, 2002). Previous estimates for the ratio of oceanic hydrothermal to continental sources for Sr in the Archean have suggested a source flux ratio of 3.1 (Kamber and Webb, 2001). In the following sections we test whether a seawater 87Sr/86Sr ratio of 0.70139 9is reasonable given the geologic constraints on ± Sr sources at 3.26 Ga.

6.1. Oceanic hydrothermal component

Using the temporal variation in 87Sr/86Sr ratio of the mantle to estimate the hydrothermal input through time (e.g., initial 87Sr/86Sr of 0.6995 for the solar system to present-day hydrothermal input of 87Sr/86Sr 0.70403; Kamber and Webb, 2001) pro- ∼ Fig. 5. A plot showing the effect of mixing hydrothermal and continental Sr given duces an 87Sr/86Sr ratio of 0.7008 at 3.26 Ga. This composition the residence time of Sr in the modern ocean (grey curve) and a residence time ∼ is consistent with estimates for mantle evolution by McCulloch that is decreased by 89% (black curve). The residence time of elements in seawater is dependent on the ratio of input/output. To test the effect of a much reduced (1994) and Kamber (2010) as well as the composition calculated residence time in the Archean we assumed all Sr input was from the continents from a direct measurement of the underlying komatiite (Jahn and and there was a net positive outflux of Sr at oceanic hydrothermal centers. We used Shih, 1974). Therefore, the Sr isotope composition of the under- the ratio of crustal abundance to relative mantle temperature at 3.2 Ga (McGovern lying komatiite is similar to the mantle, and fluids circulating and Schubert, 1989; Dhuime et al., 2012)to scale the residence time. This scaling through mantle-derived oceanic crust, would likely have had an produces a residence time that is 89% lower than in the modern oceans. This shows, 87 86 that even with a significantly reduced Sr residence time (black curve) the Archean Sr/ Sr between 0.7007 and 0.7008 at 3.2 Ga. ocean would still be isotopically homogeneous with respect to Sr. The end member isotope composition of 0.7007 is the possible hydrothermal input and 0.704 is from 6.2. Continental crust the most radiogenic barite measured as part of this study and is meant to represent the isotope composition of continental runoff. The Sr isotope composition of continental crust through time has been difficult to assess, and thus the Sr isotope composition 5.2. Strontium residence time and homogeneity of seawater of continental runoff through time has some uncertainty (Shields, 2007). A simplistic model, which is taken after Kamber and Webb An alternative explanation for the range in 87Sr/86Sr ratios mea- (2001), uses the 87Sr/86Sr ratio of BABI (0.6995) and connects to sured for barite is that they reflect Sr isotope variability in the average modern crust and its sedimentary cover (0.7119) by way marine setting in which they were deposited, which would re- of a linear evolution. With this evolution, the 87Sr/86Sr ratio of quire a relatively short Sr residence time. In the modern oceans crust at 3.26 Ga is 0.703. Even with this estimation, it is the the residence time of Sr is 2.5 m.y. (Hodell et al., 1990), which ∼ 87Sr/86Sr ratio of continental runoff that ultimately impacts the is much longer than ocean circulation time of 1500 yr (e.g., ∼ isotope composition of seawater (Shields, 2007). The Sr isotope Broecker and Peng, 1982). The long residence time of Sr rela- composition of modern river runoff is a mixture of silicate de- tive to ocean circulation time produces global ocean waters that tritus (87Sr/86Sr 0.7178), and dissolved carbonate (87Sr/86Sr are isotopically homogeneous. Previous research has shown that = = 0.7077), which mix in a ratio in favor of carbonate (Brass, 1976; even in a restricted ocean basin, chemical precipitates record the Berner and Rye, 1992 and Veizer and Mackenzie, 2003) to pro- global Sr isotope composition for seawater (Winter et al., 1997). 87 86 Some workers, however, have argued that the Sr residence time in duce a combined river runoff Sr/ Sr ratio of 0.7119 (Palmer and Archean oceans may have been shorter than today (Kamber and Edmond, 1989). Archean carbonate is rare and likely did not con- Webb, 2001). As shown in Fig. 5, only exceedingly short residence tribute as much strontium to river runoff as compared to modern times that are several orders of magnitude lower than today could continental runoff (Kamber, 2010), thus in the Archean, river runoff produce Sr isotope heterogeneity in the oceans that encompasses was likely not diluted by less radiogenic dissolved carbonate, sug- 87 86 the range of 87Sr/86Sr ratios measured for barite in this study. Us- gesting that river runoff more closely matched the Sr/ Sr ratio ing a modern Sr residence time (2.5 m.y.), ocean circulation time of silicate detritus. Therefore, with little to no carbonate to lower 87 86 would need to be very sluggish, greater than a few 100 ka, be- the Sr/ Sr ratio of continental runoff, we assume that conti- 87 86 fore Sr isotope heterogeneity could be seen on a global scale. An nental runoff, at a minimum, should have had an Sr/ Sr ratio of 0.703 at 3.2 Ga, but could be higher. exceedingly long ocean circulation time is unlikely given that mix- ∼ ing times of the oceans when continental mass was smaller should have been shorter. Alternatively, the calculations in Fig. 5 indicate 6.3. Modeling the Sr fluxes into seawater at 3.2 Ga that if ocean mixing times were similar to those of today, Sr isotope heterogeneity would require very short residence times, less Based on the mass flux model employed by Kamber (2010), than 275 kyr. The elevated 87Sr/86Sr ratios of the granular barite and using the Sr isotope composition for seawater suggested here are therefore interpreted to reflect open ocean conditions, and can- (0.70139), we calculate the Sr flux into 3.26 Ga seawater. Our re- not be explained as recording, for example, a restricted basin, as sults are based on the total flux ratio of the two primary sources of illustrated in Fig. 5. Sr into seawater, which were oceanic hydrothermal fluids and con- tinual crust (flux ratio oceanic flux/continental flux). For exam- = 6. Strontium fluxes into Archean seawater ple, using a mantle Sr isotope composition of 0.7008, and using the minimum 87Sr/86Sr ratio for continental crust (0.703), the flux ra- Our proposed 87Sr/86Sr ratio for seawater at 3.26 Ga (0.70139 tio needed to produce seawater with an 87Sr/86Sr ratio of 0.70139 ± 9) is significantly more radiogenic than contemporaneous mantle- is 2.8 or 2.3 if a 87Sr/86Sr ratio of 0.7007 is used for the mantle. 34 A.M. Satkoski et al. / Earth and Planetary Science Letters 454 (2016) 28–35

Fig. 7. Cross plot of age and 87Sr/86Sr. The carbonate chosen to help deﬁne the seawater curve proposed here has been shown to be in equilibrium in with seawater at the time of deposition (Kamber and Webb, 2001). Figure modiﬁed from Shields and Veizer (2002). Star symbol is larger than error on data.

continental exposure at this time are correct (4%; Flament et al., 2013). We suggest, however, that the global weathering rate was much higher than the modern, consistent with the inferred intense global weathering during the Archean based on the geochemistry of Archean clastic rocks (Hessler and Lowe, 2006). This view is consistent with elevated δ18Ovalues in Archean detrital zircons, which suggest that global, low-temperature alteration of continen- 87 86 Fig. 6. Cross plot of the Sr/ Sr ratio of continental runoff versus the Sr flux ratio. tal crust occurred throughout the Archean (Dhuime et al., 2012). The points represent a series of 87Sr/86Sr ratios of possible continental crust and Our results, therefore, add to the growing lines of evidence that the resultant Sr flux ratio needed to produce seawater that has a 87Sr/86Sr ratio of 0.70139. (A) Shows the results with a mantle Sr isotope composition of 0.7008 continental contributions to seawater chemistry were much higher and (B) shows the results with a mantle Sr isotope composition 0.7007. Using the in the early Archean than previously thought. isotope composition of possible crust as a proxy of continental runoff at 3.2 Ga ( 0.703; Kamber and Webb, 2001) produces a flux ratio of 2.8 (A) and 2.3 (B). ∼ 7. Evolution of Archean seawater and continental crust Using the most radiogenic barite as a proxy for continental crust (0.704) produces a flux ratio that is 4.0 (A) and 3.4 (B). The Sr flux ratio for all these scenarios is similar to the flux ratio that was calculated for 2.5 Ga seawater (Kamber and Webb, Our identification of a significantly more radiogenic seawater Sr 2001). isotope composition at 3.2 Ga (Fig. 7) is broadly consistent with the results from stratiform barite deposits (3300 150 Ma) from (Fig. 6A, B). These flux ratios are lower than predicted by Kamber ± the Indian shield that have 87Sr/86Sr ratios of 0.7018 7(Deb et ± (2010) for this time (14), but are still higher than the Phanerozoic al., 1991). Although the age and 87Sr/86Sr ratios for these deposits flux ratio (0.3–05), consistent with a significant oceanic hydrother- have high uncertainties, this 87Sr/86Sr ratio is consistent with ra- mal flux during the Archean. Our values are very similar to the diogenic seawater during this time. Thus, we propose that felsic flux ratio value predicted for 2.5 Ga seawater of 3.1 by Kamber continental crust was weathering at a rate that was significant and Webb (2001). Of note, a flux ratio of 14 calculated using a enough to move the Sr isotope composition away from a man- continental Sr isotope ratio of 0.703 and a mantle value of 0.7008 tle composition. Continental weathering affecting the Sr isotope produces seawater with a Sr isotope ratio similar to the mantle composition of Archean seawater was also proposed by Prokoph value (Kamber, 2010). To produce a fux ratio of 14 using our value et al. (2008) as a possible explanation for chemical sedimentary for seawater (0.70139) would require, for example, using a very rocks that are more radiogenic than coeval mantle. This interpre- 87 86 radiogenic Sr/ Sr ratio of 0.710 (a difference of 0.0093 between tation is consistent with 3.2 Ga clastic deposits that have dom- crust and hydrothermal input) for the composition of continental inantly felsic sources (Hessler and Lowe, 2006). The scarcity of crust at 3.26 Ga does produce a flux ratio of 14 as predicted by pre-3.2 Ga chemical sediments that faithfully represent seawater 87 86 ∼ Kamber (2010). A Sr/ Sr ratio this high for continental crust, compositions makes it difficult to track the isotope composition however, is geologically unreasonable given that the difference in of seawater back further, however, REE data from the Strelley 87 86 Sr/ Sr ratios between hydrothermal input and continental crust Pool stromatolites suggest that terrestrial weathering was affect- today is only 0.0079. We understand that the isotope composition ing seawater chemistry as far back as 3.45 Ga (van Kranendonk et of past continental flux is poorly constrained (Shields, 2007), but al., 2003). The nature of weathering at 3.45 Ga was likely mafic as described in detail in Section 6.2, we use 0.703 as a minimum lithologies (van Kranendonk et al., 2003), thus we propose that ratio in the flux calculation. As described above, a continental flux the rise in felsic continental weathering happened between 3.45 with an isotope ratio higher than 0.703 will reduce the amount of and 3.2 Ga (Fig. 7). This change in the nature of continental flux needed to produce seawater with an isotope ratio of 0.70139. weathering fluxes to the oceans at 3.2 Ga is broadly consis- 87 86 ∼ For example, increasing the Sr/ Sr value for continental crust tent with the timing observed in the increase in abundance of to that of the most radiogenic barite measured ( 0.704) would high-δ18Odetrital zircons (Dhuime et al., 2012), the appearance of ∼ raise to flux ratio needed to produce a seawater isotope compo- high-187Os/188Os eclogitic diamonds at 3 Ga (Shirey and Richard- ∼ sition of 0.70139 to between 4.0 and 3.4 (Fig. 6A, B). Flux ratio son, 2011), and 3.2 Ga clastic deposits that have dominantly felsic ratios between 2.8–4.0 may seem low given the predictions of low sources (Hessler and Lowe, 2006). These observations could imply A.M. Satkoski et al. / Earth and Planetary Science Letters 454 (2016) 28–35 35 a change in crustal dynamics and mark the onset of plate tecton- Mesoarchean Fig Tree Group, Barberton Mountainland, South Africa. In: Trans. ics, continental collision, uplift, and erosion. The elevated seawater of the V South American Symp. on Isotope Geology, pp. 258–262, contrib. 311. 87Sr/86Sr ratio we infer at 3.2 Ga, relative to the previous seawa- Heinrichs, T.K., Reimer, T.O., 1977. A sedimentary barite deposit from the Archean Fig Tree Group of the Barberton mountain land (South Africa). Econ. Geol. 72, ter evolution curves of Shields and Veizer (2002) and Flament et 1426–1441. al. (2013), supports a much earlier rise in felsic continental weath- Hessler, A.M., Lowe, D.R., 2006. Weathering and sediment generation in the ering at >3.2 Ga, and by extension, a larger volume of emergent Archean: an integrated study of the evolution of siliciclastic sedimentary rocks continental mass. In a recent data compilation for the evolution of the 3.2 Ga Moodies Group, Barberton Greenstone Belt, South Africa. Precam- brian Res. 151, 185–210. of Rb/Sr ratios in igneous rocks, Dhuime et al. (2015) showed that Hodell, D.A., Mead, G.A., Mueller, P.A., 1990. Variation in the strontium isotopic com- Rb/Sr ratios began a consistent increase in igneous rocks from 3.2 position of seawater (8 Ma to present): implications for chemical weathering Ga to the end of the Precambrian, and our results suggest that rates and dissolved fluxes to the oceans. Chem. Geol. 80, 291–307. such a signal may be detected in a revised Sr seawater curve for Huston, D.L., Logan, G.A., 2004. Barite, BIFs and bugs: evidence for the evolution of the Archean. the Earth’s early hydrosphere. Earth Planet. Sci. Lett. 220, 41–55. Jahn, B., Shih, C., 1974. On the age of the Onverwacht Group, Swaziland Sequence, Acknowledgements South Africa. Geochim. Cosmochim. Acta 38, 873–885. Jewell, P.W., 2000. Bedded barite in the geological record. In: Glenn, C.R., Prevot- Lucas, L., Lucas, J. (Eds.), Marine Authigenesis: From Global to Microbial. In: We thank two anonymous reviewers for constructive reviews SEPM Spec. Publ., vol. 66, pp. 147–161. of the manuscript, and Prof. Michael Bickle for editorial handling Kamber, B.S., 2010. Archean mafic–ultramafic volcanic landmasses and their effect and comments. This study was funded by the NASA Astrobiology on ocean–atmosphere chemistry. Chem. Geol. 274, 19–28. Institute and NSF grant 1523697. The contribution by MC was car- Kamber, B.S., Webb, G.E., 2001. The geochemistry of late Archean microbial carbon- ried out at the Jet Propulsion Laboratory (JPL), California Institute ate: implications for ocean chemistry and continental erosion history. Geochim. Cosmochim. Acta 65, 2509–2525. of Technology, under contract with the National Aeronautical and Kasting, J.F., 2014. Atmospheric composition of Hadean–early Archean Earth: the Space Administration (NASA). We thank Bethany Theiling for the S importance of CO. In: Shaw, G.H. (Ed.), Earth’s Early Atmosphere and Surface En- and O isotope analyses. vironment. In: Geological Society of America Special Paper, vol. 504, pp. 19–28. Kröner, A., Byerly, G.R., Lowe, D.R., 1991. Chronology of early Archaean granite- Appendix A. Supplementary material greenstone evolution in the Barberton Mountain Land, South Africa, based on precise dating by single zircon evaporation. Earth Planet. Sci. Lett. 103, 41–54. Supplementary material related to this article can be found on- Lowe, D.R., 2013. Crustal fracturing and chert dike formation triggered by large meteorite impacts, ca. 3.260 Ga, Barberton greenstone belt, South Africa. Geol. Soc. line at http://dx.doi.org/10.1016/j.epsl.2016.08.032. Am. Bull. 125, 894–912. Lowe, D.R., Nocita, B.W., 1999. Foreland basin sedimentation in the Mapepe Forma- References tion, southern-facies Fig Tree Group. In: Lowe, D.R., Byerly, G.R. (Eds.), Geologic Evolution of the Barberton Greenstone Belt, South Africa. In: Geological Society

Bao, H., 2015. Sulfate: a time capsule for Earth’s O2, O3, and H2O. Chem. Geol. 395, of America Special Paper, vol. 329, pp. 233–258. 87 86 108–118. McCulloch, M.T., 1994. Primitive Sr/ Sr from an Archean barite and conjecture on Bao, H., Rumble III, D., Lowe, D.R., 2007. The five stable isotope compositions of Fig the Earths age and origin. Earth Planet. Sci. Lett. 126, 1–13. Tree barites: implications on the sulfur cycle in ca. 3.2 Ga oceans. Geochim. McGovern, P.J., Schubert, G., 1989. Thermal evolution of the Earth: effects of volatile Cosmochim. Acta 71, 4868–4879. exchange between atmosphere and interior. Earth Planet. Sci. Lett. 96, 27–37. Beaulieu, E., Godderis, Y., Donnadieu, Y., Labat, D., Roelandt, C., 2012. High sensitiv- Palmer, M.R., Edmond, J.M., 1989. Strontium isotope budget of the modern ocean. ity of the continental-weathering carbon dioxide to future climate change. Nat. Earth Planet. Sci. Lett. 92, 11–26. Clim. Change 2, 346–349. Perry, E.C., Monster, J., Reimer, T., 1971. Sulfur isotopes in Swaziland system barites Berner, R.A., Rye, D.M., 1992. Calculation of the Phanerozoic strontium isotope and the evolution of the Earth’s atmosphere. Science 171, 1015–1016. record of the oceans from a carbon cycle model. Am. J. Sci. 292, 136–148. Philippot, P., Van Zuilen, M., Lepot, K., Thomazo, C., Farquhar, J., Van Kranendonk, Blake, R.E., Chang, S., Lepland, A., 2010. Phosphate oxygen isotopic evidence for a M.J., 2007. Early Archaean microorganisms preferred elemental sulfur, not sul- temperate and biologically active Archean ocean. Nature 464, 1029–1033. fate. Science 317, 1534–1537. Brass, G.W., 1976. The variation of the marine 87Sr/86Sr ratio during Phanero- Prokoph, A., Shields, G.A., Veizer, J., 2008. Compilation and time-series analysis of a 18 13 87 86 34 zoic time: interpretation using a flux model. Geochim. Cosmochim. Acta 40, marine carbonate δ O, δ C, Sr/ Sr and δ Sdatabase through Earth history. 721–730. Earth-Sci. Rev. 87, 113–133. Breit, G.N., Simmons, E.C., Goldhaber, M.B., 1985. Dissolution of barite for the anal- Reimer, T.O., 1980. Archean sedimentary baryte deposits of the Swaziland Super- ysis of strontium isotopes and other chemical and isotopic variations using group (Barberton Mountain Land, South Africa). Precambrian Res. 12, 393–410. aqueous sodium carbonate. Chem. Geol. 52, 333–336. Roerdink, D.L., Mason, P.R.D., Farquhar, J., Reimer, T., 2012. Multiple sulfur isotopes in Broecker, W.S., Peng, T.-H., 1982. Tracers in the Sea. Eldigio Press, Palisades, NY. the Paleoarchean barites identify an important role for microbial sulfate reduc- Canfield, D.E., Farquhar, J., 2009. Animal evolution, bioturbation, and the sulfate con- tion in the early marine environment. Earth Planet. Sci. Lett. 331–332, 177–186. centration of the oceans. Proc. Natl. Acad. Sci. USA 106, 8123–8127. Shields, G.A., 2007. A normalized seawater strontium isotope curve: possible impli- Chiba, H., Sakai, H., 1985. Oxygen isotope exchange rate between dissolved sul- cations for Neoproterozoic–Cambrian weathering rates and the further oxygena- fate and water and hydrothermal temperatures. Geochim. Cosmochim. Acta 49, tion of the Earth. eEarth 2, 35–42. 993–1000. Shields, G.A., Veizer, J., 2002. Precambrian marine carbonate isotope database: ver- Clemens, J.D., Yearron, L.M., Stevens, G., 2006. Barberton (South Africa) TTG magmas: sion 1.1. Geochem. Geophys. Geosyst. 3, 1–12. geochemical and experimental constraints on source-rock petrology, pressure of Shirey, S.B., Richardson, S.H., 2011. Start of the Wilson Cycle at 3 Ga Shown by Dia- formation and tectonic setting. Precambrian Res. 151, 53–78. monds from Subcontinental Mantle. Science 333, 434–436. Deb, M., Hoefs, J., Baumann, A., 1991. Isotopic composition of two Precambrian strat- Ushikubo, T., Kita, N.T., Cavosie, A.J., Wilde, S.A., Rudnick, R.L., Valley, J.W., 2008. iform barite deposits from the Indian shield. Geochim. Cosmochim. Acta 55, Lithium in Jack Hills zircons: evidence for extensive weathering of Earth’s earli- 303–308. est crust. Earth Planet. Sci. Lett. 272, 666–676. Dhuime, B., Hawkesworth, C.J., Cawood, P.A., Storey, C.D., 2012. A change in geody- Van Kranendonk, M.J., Webb, G.E., Kamber, B.S., 2003. Geological and trace element namics of continental growth 3 billion years ago. Science 335, 1334–1336. evidence for a marine sedimentary environment of deposition and biogenicity Dhuime, B., Wuestefeld, A., Hawkesworth, C.J., 2015. Emergence of modern conti- of 3.45 Ga stromatolitic carbonates in the Pilbara Craton, and support for a re- nental crust about 3 billion years ago. Nat. Geosci. 8, 552–555. ducing Archaean ocean. Geobiology 1, 91–108. Eriksson, K., Soegaard, K., 1985. The petrology and geochemistry of Archean and Veizer, J., Mackenzie, F.T., 2003. Evolution of sedimentary rocks. In: Treatise on Geo- early Proterozoic sediments: implications for crustal compositions and surface chemistry, pp. 369–407. edited by: F.T., volume 7, Sediments, diagenesis, and processes. In: Laajoki, K., Paakkola, J. (Eds.), Proterozoic Exogenic Processes and sedimentary rocks. Related Metallogeny, vol. 331. Geological Society of Finland, pp. 9–30. Winter, B.L., Johnson, C.M., Clark, D.L., 1997. Strontium, neodymium and lead isotope Flament, N., Cotice, N., Rey, P.F., 2013. The evolution of the 87Sr/86Sr of marine car- variations of authigenic and silicate sediment components from the Late Ceno- bonates does not constrain continental growth. Precambrian Res. 229, 177–188. zoic Arctic Ocean: implications for provenance and the source of trace metals in Griffith, E.M., Paytan, A., 2012. Barite in the ocean – occurrence, geochemistry and seawater. Geochim. Cosmochim. Acta 19, 4181–4200. palaeoceanography applications. Sedimentology 59, 1817–1835. Zeh, A., Gerdes, A., Heubeck, C., 2013. U–Pb and Hf isotope data of detrital zircons Gutzmer, J., Banks, D., de Kock, M.O., McClung, C.R., Strauss, H., Mezger, K., 2006. from the Barberton Greenstone Belt: constraints on provenance and Archean The origin and paleoenvironmental significance of stratabounds barite from the crustal evolution. J. Geol. Soc. Lond. 170, 215–223.