Coupling of the atmosphere and sediment melts across the Archean-Proterozoic transition

The Archean-Proterozoic transition marks a time of fundamental geologic, biologic, and atmospheric changes to the Earth system, including oxygenation of the atmosphere (termed the Great Oxygenation Event; GOE), and the emergence of continents above sea-level. The impacts of the GOE on Earth’s surface environment are imprinted on the geologic record, including the attenuation of mass-independent fractionation of sulfur isotopes (S-MIF). Temporally overlapping geologic and geochemical observations (e.g. a change in oxygen isotope ratio of sediment melts) imply the widespread subaerial emergence of continents was coeval with atmospheric oxygenation. Here we present triple sulfur isotope ratios in pyrite and oxygen isotope ratios in garnet and zircon in a global suite of Archean and Proterozoic sediment-derived granitoids. These crustal melts record an increase in average 18O/16O isotope ratio and a disappearance of S-MIF in the Paleoproterozoic. The coupled behaviour of sulfur and oxygen isotope signatures imply a potential causal link between the emergence of continents and atmospheric oxygenation at ~2.3 Ga.


Introduction
The oxygenation of the Earth's atmosphere and oceans irreversibly changed many major biogeochemical cycles (e.g. Fe, S, Mn) 1 and provided the base for a highly e cient aerobic metabolism that allowed the development of complex life 2 . During the Great Oxygenation Event (GOE) (c. 2.3 Ga) atmospheric oxygen increased from <0.001% of the present atmospheric level (PAL) 3 to 10-40% PAL 4 , or perhaps much higher approaching 100% PAL 5,and references therein . The mechanism that led to atmospheric oxygenation remains controversial. Proposed scenarios for oxygenation include 1) an increase in O 2 production (i.e. through the emergence of oxygenic photosynthesis) 6 , 2) a decrease in O 2 consumption (e.g. through changing redox state of volcanic gases, increased burial of organic carbon, or decreased pyrite weathering) 7,8 , and 3) a combination of both processes (through enhanced oxygenic photosynthesis combined with increased carbon burial) 9 . One source of dispute is the timing of the invention of oxygenic photosynthesis; suggestions range from >3.7 Ga 10 to immediately preceding atmospheric oxygenation 6 . Previous studies imply that the widespread emergence of continents above sea-level is temporally correlated with atmospheric oxygenation 11 . Furthermore, it has been proposed that the subaerial emergence of continents may have led to a ux of life-essential nutrients into the ocean supporting a boost in photosynthetic activity 12 .
Multiple sulfur isotopic signatures are a sensitive tracer for atmospheric oxygen levels 13 . Some sul des and sulfates in metasedimentary rocks deposited prior to the GOE display mass-independent fractionation of sulfur (S-MIF), whereas those deposited after the GOE display almost wholly massdependent fractionation (S-MDF) 13,14 . It has been postulated that S-MIF was generated in the atmosphere through ultraviolet photolysis of gas molecules 15 . The establishment of an ozone shield as a consequence of atmospheric oxygenation led to blocking of UV radiation and attenuation of the photolysis of volcanic sulfur species 13 . In addition, enhanced oxidative weathering of sul des and the formation of oceanic sulfates stimulated the activity of sulfur-metabolizing bacteria supporting the generation of S-MDF signatures 16 . Recent studies show that uctuations in atmospheric oxygen level are also captured in the igneous rock record; namely, through recycling of S-MIF in the crust 17,18 , and a change in oxygen fugacity of strongly peraluminous granites 19 .
Broadly coeval with atmospheric oxygenation, the average oxygen isotope ratio of global felsic magmas (recorded by zircon δ 18 O) increases 20 . The oxygen isotopic composition of a magma is sensitive to the recycling of supracrustal material 21 25 . Therefore, a change in sediment oxygen isotope composition potentially associated with the subaerial emergence of continents seems to be the likely driver for the Paleoproterozoic increase in average zircon δ 18 O 25 .
We explore a potential link between the emergence of continents above sea-level and atmospheric oxygenation through a coupled analysis of proxies for atmospheric oxygen level and sedimentary recycling in Archean to Mesoproterozoic sediment-derived granitoids (i.e. granitoids that partially or wholly derive from the partial melting of metasediments). Here we present pyrite multiple sulfur isotope ratios in tandem with zircon and garnet oxygen isotope ratios for a global sample set. The combination of these proxies allows us to provide new insights into the coupled behaviour of geodynamic, biogenic, and atmospheric evolution.

Bulk-rock geochemistry and mineralogy
Samples include granitoids comprised of quartz + alkali feldspar + plagioclase ± biotite in varying proportions, and contain one or more peraluminous indicator minerals, such as garnet or muscovite. The samples for which bulk-rock geochemical data is available are strongly peraluminous with an aluminium saturation index (ASI) ≥ 1.1 (de ned as molecular Al/[Ca -1.67P + Na + K]). For comparison ACNK values (de ned as molecular Al/[Ca + Na + K]) are also reported in Table 2. With the exception of sample 15K-2, ASI and ACNK yield identical values to the rst decimal place. The mineralogy of all samples is summarized in Table 1. Bulk rock major element concentrations are given in Table 2. Thin section photomicrographs can be found in supplementary Figure A4.

Geochronology
Magmatic crystallization ages of the 30 sediment-derived granitoids in this study are Neoarchean to Mesoproterozoic, ranging from 2664 ± 45 Ma to 1447 ± 50 Ma. For nine of these samples the magmatic crystallization age was determined in this study (as zircon concordia, upper intercept, or weighted mean 207 Pb*/ 206 Pb* ages). For nine samples zircon U-Pb SIMS ages were determined in previous studies 25 .
For the remaining 12 samples for which no or only metamict zircon was extracted, preferred ages use robust published dates from the same batholith or are estimates based on the age of proximal magmatism. A detailed description of the geochronology is given in Appendix A; a summary is given in Table 3. Single-spot zircon U-Pb results can be found in supplementary  10.24‰ in garnet, and from 7.19‰ to 9.30‰ combining the data of both zircon and garnet. Equilibrium fractionation of oxygen isotopes between zircon and almandine-rich garnet is small at temperatures typical for granitoid melts (<0.1‰ at temperatures >650 °C) 27 . The garnet-zircon pairs of all samples indicate oxygen isotopic equilibrium ( Figure 1). Two samples (17FIN02 and 17FIN04A) yield heterogeneous single spot zircon δ 18 O values (2σ > 3‰) and are interpreted to re ect secondary signatures 28 . This is further supported by CL images revealing that some areas in some zircon grains are affected by metamictization (supplementary Figure A2). Therefore, the oxygen isotopic composition recorded by garnet from these two samples provides the best estimate of their parental magma δ 18 O. Data tables with single spot O isotopic data are given in the supplementary Table D1, a summary of weighted means is given in Table 3.

Sulfur isotope geochemistry
Pyrite grains are euhedral to sub-euhedral, chemically homogenous, and largely free of inclusions and intergrown phases. BSE images of representative pyrite grains for each sample are given in the supplementary Figure A3. Pyrite δ 34 S values range from -13.33 ± 0.36‰ to 9.72 ± 0.99‰, but are mostly (11 out of 13 samples) between -4‰ and 4‰. Single spot pyrite δ 34 S and ∆ 33 S values cluster tightly around discrete values for each sample, and seemingly de ne single populations ( Figure 2). Four samples show pyrite ∆ 33 S outside of the S-MDF range ( Figure 2). These samples include three ~2.7 Ga granites from the Superior province that exhibit positive ∆ 33 S values of 0.13 ± 0.06‰ to 0.18 ± 0.05 ‰, and a ~2.5 Ga granite from the North China Craton with negative ∆ 33 S value of -0.29 ± 0.12‰. Data tables with single spot sulfur isotopic data are given in the supplementary Table C1, a summary of weighted means is given in Table 3.

Discussion
The presence of aluminous mineral phases and/or ASI ≥ 1.1 strongly suggest that the studied granitoids were derived from the partial melting of metasedimentary protoliths 29,30 . The partial or entire derivation of these granitoids from metasedimentary protoliths is in accord with previous interpretations of the regional geology [31][32][33][34] . Zircon and garnet tend to preserve a record of the oxygen isotope composition of their parental melt due to slow intracrystalline diffusion rates of oxygen in these minerals 35,36 . Given the small equilibrium fractionation (smaller than the analytical uncertainty) of oxygen isotopes between almandine-rich garnet and zircon at temperatures typical for granitoid melts 27 37 . It has been proposed that interaction of a melt with hydrothermal meteoric water -a mechanism that requires emergent land area -may be the process responsible for such low δ 18 O melts 37 . The emergence of continents at ~2.4 Ga is supported by changes to geochemical proxies at this time, such as an increase in seawater 87 Sr/ 86 Sr 38 , a decrease in shale ∆ 17 O 24,39 , as well as an increase in subaerial large igneous province volcanism 11 .
Pyrite-bearing sediment-derived granitoids with crystallization ages >2.3 Ga yield non-zero ∆ 33 S values, whereas those younger than 2.3 Ga uniformly show ∆ 33 S of 0‰. Although the ∆ 33 S values of the >2.3 Ga granitoids are small compared to those in pre-GOE sedimentary rocks (up to ~12‰) 14 , the magnitude of non-zero ∆ 33 S values observed here is too large to be the result of S-MDF processes alone ( Figure 2).
Hence, the S-MIF signatures in the >2.3 Ga pyrite-bearing granitoids record the recycling of sedimentary sulfur species formed under the anoxic pre-GOE atmosphere, as has been demonstrated for S-MIF carrying strongly peraluminous granites in previous studies 18 . Three of the >2.3 Ga granitoids of this study show positive ∆ 33 S values, one sample shows a negative ∆ 33 S value. The pre-GOE pyrite record is skewed towards positive ∆ 33 S values 14 . Positive ∆ 33 S anomalies are commonly found in Archean sedimentary pyrite that may have formed from reduced sulfur species (e.g. S 8 aerosols) produced through photodissociation in the oxygen-poor atmosphere 13,40 . Based on the dominance of negative ∆ 33 S values in Archean barites, it has been posited that oceanic sulfates (formed from oxidized sulfur species produced through photodissociation; e.g. SO 4 aerosols) carry the complementary negative ∆ 33 S signatures required by isotopic mass balance 13 . However, recent studies report positive ∆ 33 S anomalies in Archean oceanic sulfate 41 . The inconsistent sulfur isotopic budget is a yet unsolved scienti c problem, and the complementary negative ∆ 33 S reservoir remains cryptic.
The low δ 34 S value (~ -13‰) of pyrite in sample SP-17-43 from the Wenasaga Lake batholith, Superior province, could derive from the contribution of organic sulfur. Mircobial sulfate reduction commonly produces sul des that are strongly depleted in the heavy 34 S isotope 42 . The S-MIF signature in this sample (∆ 33 S = 0.16 ± 0.04‰) indicates atmospheric in uence. Therefore, the recorded sulfur isotopic composition of this sample may be the result of mixing between two reservoirs (i.e. atmospheric and microbial sulfur). A similar scenario has been suggested to be responsible for the negative δ 34  The implication of our study is that the subaerial emergence of continents constitutes a potential driver for ecological changes that fuelled oxygenic photosynthesis, ultimately leading to a change in redox state of the Earth's atmosphere and oceans.

Sample context
Samples of this study derive from various localities within the Superior, North China, West African, and East European Cratons, and the Yavapai province (USA). A list of all samples including location is given in Table 1.

Superior Craton
The Superior Craton forms the Archean core of the Canadian Shield and can be divided into the Western and Eastern Superior Province, which are further subdivided into 17 distinct tectonic terranes 31 .

Ukrainian Shield, East European Craton
The Ukrainian Shield is a region of exposed Archean and Proterozoic crust within Samartia in the southwestern part of the East European Craton. The Ukrainian Shield is comprised by several tectonic blocks separated by suture zones described in detail by 59  This complex, referred to as the late Svecofennian granite-migmatite zone, includes the Sulkava, West Uusuma, and Turku areas 61 . Granitic material is abundant in the Turku area. Mostly, these magmas occur as garnet and cordierite bearing leucosome in magmatic metapelites 34 .

Yavapai province
The Yavapai province (or Colorado province) south of the Wyoming Craton comprises ~1.79-1.66 Ga volcanic-plutonic suites and sediments that are interpreted to have formed in a convergent margin setting 62 . These rocks experienced multiple deformational episodes associated with metamorphism and plutonism between 1.71 and 1.62 Ga described in detail by Hoffman, (1988). The deformational episodes were followed by two pulses of calc-alkaline to alkaline magmatism at 1.50-1.      Zircon δ18O vs garnet δ18O color-coded by crystallization age of sediment-derived igneous rocks.
Isotherms at 650°C and 1200°C are after 27. Oxygen isotope data is shown as weighted averages. Single spot results of O isotope analysis are given in the appendix. Error bars are shown at 2σ level. Note that large error bars for zircon δ18O are due to heterogeneity in these samples interpreted to be related to secondary processes (discussed in the text).

Figure 2
Single spot δ34S vs. ∆33S of sediment-derived igneous rocks. Error bars for ∆33S are shown at 1σ level.
Errors of δ34S at 2σ level are smaller than the symbols. Grey area marks the range of ∆33S values that can be produced through mass-dependent fractionation processes 83. Single spot results of sulfur isotope analysis are given in the appendix.

Figure 3
Oxygen and sulfur isotopic data vs. crystallization age of sediment-derived igneous rocks. Timing of atmospheric oxygenation 84 is shown as blue bar. Legend on the right applies to both parts of the gure. (A) Zircon and garnet δ18O vs. crystallization age of sediment-derived igneous rocks. Oxygen isotope data is shown as weighted averages. Grey bars mark average δ18O (as recorded by zircon and garnet) pre-and post-2.3 Ga, respectively. Single spot results of O isotope analysis are given in the appendix.
Error bars for oxygen isotope data and age are 2σ. Zircon oxygen isotope data from Suomussalmi and Helanshan are from 1)85, and 2)86.