FEATURE ARTICLE Evolutio n of t he Global Carbo n Cycle a nd 10.1029/2018 GB006061 Cli mate Regulation on Earth

T. T. Isso n 1, 2 , N. J. Pla navsky1 , L. A. Cooga n3 , E. M. Ste wart1 , J. J. Ag ue1 , E. W. Bolto n1 , S. Z h a n g1 , N. R. Mc Kenzie 4 , and L. R. Ku mp5

1 Depart ment of Geology and Geophysics, Yale University, Ne w Haven, CT, US A, 2 Sc hool of Scie nce, U niversity of Waikato (Tauranga), Tauranga, Ne w Zealand, 3 Sc hool of Eart h a nd Ocea n Scie nces, U niversity of Victoria, Victoria, British Colu mbia, Canada, 4 Depart ment of Earth Sciences, University of Hong Kong, Hong Kong, 5 Depart ment of Geosciences, Pennsylvania State University, University Park, P A, US A

Key Poi nts: • Long ter m carbon sources and sinks are li kely larger t ha n traditio nal A b st r a ct The existence of stabilizing feedbacks within Earth's cli mate syste m is generally thought to e n visi o n e d be necessary for t he persiste nce of liquid water a nd life. Over t he course of Eart h's history, Eart h's • There is signi fi cant silicate at mosp heric co mpositio n appears to have adjusted to t he gradual i ncrease i n solar lu mi nosity, resulti ng i n weat heri ng i n t he mari ne as well as terrestrial setti ngs persistently habitable surface te mperatures. With li mited exceptions, the Earth syste m has been observed • Ocean oxygenation and evolution of to recover rapidly fro m pulsed cli matic perturbations. Carbon dioxide ( C O2 ) regulatio n via negative a biotic Si cycle forced a drop i n feedbacks within the coupled global carbon‐silica cycles are classically vie wed as the main processes reverse weathering rates and an increase in marine weathering rates giving rise to cli mate stability on Earth. Here we revie w the long‐ter m global carbon cycle budget, and ho w the processes modulating Earth's cli mate syste m have evolved over ti me. Speci fically, we focus on the relative roles t hat s hifts i n carbo n sources and sinks have played i n drivi ng long‐ter m c ha nges i n at mospheric p C O . We make the case that marine processes are an i mportant co mponent of the canonical Correspondence to: 2 T. T. Isso n a nd N. J. Pla navsky, silicate weathering feedback, and have played a much more i mportant role in pC O 2 regulation than terry.isson @ waikato.ac.nz; traditionally i magined. Notably, geoche mical evidence indicate that the weathering of marine sedi ments noah.planavsky @yale.edu and off‐axis basalt alteration act as major carbon sinks. Ho wever, this sink was potentially da mpened d uring Earth's early history w he n oceans had higher levels of dissolved silicon (Si), iron ( Fe), a nd Cit ati o n: magnesiu m ( Mg), and instead likely fostered more extensive carbon recycling within the ocean ‐at mosphere Isso n, T. T., Pla navsky, N. J., Cooga n, syste m via reverse weathering— that in turn acted to elevate ocean‐at mosphere C O l e v els. L. A., St e w art, E. M., A g u e, J. J., B olt o n, 2 E. W., et al. (2020). Evol utio n of t he global carbon cycle and cli mate regulation on earth. Global Biogeoche mical Cycles ,34, 1. I ntrod uctio n e2018 G B006061. https://doi.org/ 10.1029/2018 GB006061 Liquid water is esse ntial for life as we k no w it ( Dole, 1964). Alt houg h water is prese nt o n multiple bodies within our solar syste m, only one planet — Earth — has sustained liquid water at the surface for the majority Received 4 F E B 2019 of its history ( Mojzsis et al., 2001; Wilde et al., 2001). Despite large c ha nges i n solar l u mi nosity ( Go ug h, 1981; Accepted 20 N O V 2019 Hoyle, 1957), ocean che mistry, and the rock cycle over ti me ( Berner, 2004; Holland, 1984, 2002; Mackenzie Accepted article online 30 D E C 2019 & Garrels, 1971), processes within Earth's syste m have naturally regulated greenhouse gas levels so as to mai ntai n relatively i nvaria nt pla netary te mperat ures ( Ber ner et al., 1983; Hart, 1978; Kasti ng, 1987; Saga n & M ulle n, 1972). T his cli mate stability has allo wed for t he persiste nt i n habitatio n a nd proliferatio n of co m- plex life over billio ns of years, a nd is typically ta ke n as stro ng evide nce for t he existe nce of stabilizi ng feed- backs on Earth (e.g., Berner & Caldeira, 1997; Kasting, 2019; Lovelock & Whit field, 1982). Classically, cli mate regulation on Earth is vie wed to be tied fore most to the regulation of at mospheric carbon dioxide

(CO2 ) levels, via processes s uc h as terrestrial silicate weat heri ng t hat co ntroll t he rate of carbo n re moval fro m the ocean‐at mosphere syste m. Within this syste m, steady‐state and transient variations to a range of

carbon sources and sinks have been proposed to drive major changes in pC O 2 and thus Earth's cli mate. Collectively, the robustness and character of Earth's cli mate regulating mechanis ms have been proposed to play a sig nifi ca nt role i n gover ni ng its lo ng‐ter m habitable life spa n ( Caldeira & Kasti ng, 1992; Li et al., 2009; Lovelock & Whitfi eld, 1982). In this light, a mechanistic understanding of Earth's cli mate feedbacks can help guide our vie ws on planetary habitability beyond our solar syste m. Here, we revie w key co ncepts a nd c ha nges to traditio nal vie ws of t he lo ng ‐ter m carbo n cycle (sectio n 2), its budget (section 3), and its evolution through ti me (section 4). Rather than providing an overvie w of all

©2019. A merican Geophysical Union. aspects of the long‐ter m cli mate controls, we revie w the main processes leading to cli mate regulation and All Rig hts Reserved. focus our discussio n o n pote ntial deviatio ns fro m t he sta ndard vie w (e.g., Ber ner, 2004).

ISS O N ET AL. 1 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

Fig ure 1. Si mplifi ed cartoon illustrating the inorganic portion of the global long‐ter m carbon cycle. Red arro ws indicate carbon entering and leaving the ocean‐ at mosphere syste m. Blue and green arro ws indicate weathering and reverse weathering processes respectively, that alter the species of carbon (bet ween C O 2 and carbonate alkalinity) within the ocean‐at mosphere syste m.

2. Cli mate Regulation Through the Carbon Cycle 2.1. Silicate Weat heri ng

Carbon enters the ocean ‐at mosphere reservoir through solid Earth degassing of C O 2 a n d is l ost t hr o u g h t h e sequestration of carbon as carbonate ( X C O 3 , here a nd else w here X represe nts a divale nt catio n, typically 2 + 2 + 2 + C a , M g , or F e ) or bio mass ( C H2 O) (Figure 1) ( Cha mberlin, 1899; Högbo m, 1894; Urey, 1952). The path ways leading to the release and dra wdo wn of carbon have long been recognized and can be si mply expressed as t wo f u nda me ntal reactio ns t hat hig hlig ht t he i norga nic (eq uatio n (1)) a nd orga nic (eq uatio n (2)) portio ns of t he lo ng‐ter m carbo n cycle.

X Si O 3 þ C O 2 ↔ X C O 3 þ Si O 2 ( 1)

T his so ‐called Urey eq uatio n (eq uatio n (1)) is t he si mplest descriptio n of t he i norga nic carbo n cycle. I n t he

for ward (left to rig ht) directio n, t he weat heri ng of a silicate mi neral ( XSi O3 ) is co upled to t he re moval of car- bo n as carbo nate a nd silico n as c hert (Si O 2 ). T he reverse directio n (rig ht to left) ill ustrates C O2 degassi ng as a result of ther mal decarbonation reactions during meta morphis m.

CO 2 þ H 2 O ↔ C H 2 O þ O 2 ( 2)

Eq uatio n (2) represe nts t he orga nic portio n of t he carbo n cycle. Here, t he for ward (left to rig ht) directio n

describes carbon fixation, releasing oxygen ( O 2 ) i n t he process, w hile t he reverse (rig ht to left) acco u nts for the oxidation of organic matter. Although water ( H2 O) is not t he o nly electro n do nor (i.e., a noxyge nic p hotosy nt hesis reactio ns exist), a nd si milarly, oxyge n not t he o nly oxidizi ng age nt available, t hese represe nt the predo minant path ways of organic matter synthesis and re mineralization in the modern oceans. It is i mporta nt to note t hat t he orga nic portio n of t he global carbo n cycle is classically not vie wed to play a s ubsta ntial role i n reg ulati ng cli mate stability give n t hat oxyge n so urces a nd si nks are likely to bala nce eac h

other due to the presence of strong negative feedbacks on at mospheric O 2 (e.g., Berner, 2004; Laakso & Schrag, 2014; Ozaki et al., 2019). For instance, enhanced organic carbon production elevates free O 2 a n d increases organic matter oxidation. Ho wever, there are exceptions to this vie w (e.g., France‐Lanord & Derry, 1997; Galy et al., 2007; Galy et al., 2010). T he size of t he ocea n ‐at mosp here carbo n reservoir ( Table 1) at a ny give n i nterval i n Eart h's history is co n-

trolled by a bala nce bet wee n rates of carbo n i nput ( Fi n), and the ease with which carbon leaves the syste m (i.e., t he effi cie ncy of carbo n sequestratio n for a give n Eart h state). All else bei ng equal, a n i ncrease i n

ISS O N ET AL. 2 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

T a bl e 1 degassing rates or an Earth environ ment more resistant to carbon re moval Modern Earth Surface Carbon Reservoirs (decrease weatherability or increase reverse weathering) will lead to an 1 8 Carbon reservoirs M ass ( 1 0 m ol) increase in at mospheric pC O 2 ( Fig ures 1–3). Solid Eart h processes (e.g., O c e a n 2. 8 arc volcanis m, mid‐ocean ridge spreading rates, and meta morphic decar- At m os p h er e 0. 0 6 bonation) govern the rates of ne w carbon being introduced (discussed in S oil 0. 3 section 3.1). Carbon re moval, on the other hand, involves t wo steps, fi rst Terrestrial biosp here 0. 0 6 − 2 − the conversion of C O2 to carbonate alkalinity ( H C O3 a n d C O 3 ), a n d Marine biosphere 0. 0 5 second the re moval of carbonate alkalinity as carbonate rock ( X C O ). Upper co nti ne ntal crust 6, 2 5 0 3 ( Carbonate C in rocks) ( 5, 0 0 0) The ease to which C O 2 is co nverted to carbo nate alkali nity is deter mi ned ( Organic C in rocks) ( 1, 2 5 0) by Earth's susceptibility to silicate weathering, a property co m monly Note . Source: Berner, 2004. referred to as the “weatherability ” of Earth's surface environ ment (Figures 2 and 3). Although the concept of weatherability is most co m- mo nly applied o nly to terrestrial silicate weat heri ng, silicate weat heri ng in the marine real m (in marine sedi ments and oceanic crust) ought to be included when referring to the weatherability of the Earth surface syste m as a whole. Nu merous factors have been proposed to control Eart h's surface weat herability i n terrestrial e nviro n me nts, i ncludi ng crustal co mpositio n (lit hology), tec- to nics (e.g., uplift rates, topograp hy), co nti ne ntal co nfi g uratio n (paleogeograp hy), a nd hydrology a nd biolo- gical alteration (effect of land plants and soil bio mass) ( Ku mp & Arthur, 1997). On the other hand, the weatherability of the marine environ ment depends on the rate of for mation of ne w oceanic crust, its bulk co mposition, abyssal sedi mentation rate, and the saturation state of sea water with respect to silicate p hases— all of w hic h have likely evolved over t he course of Eart h's history.

Fig ure 2. Key principles of the global carbon cycle and at mospheric C O 2 . T he effect of (a) o utgassi ng, (b) stre ngt h of t he silicate weat heri ng feedback, (c) orga nic carbo n a nd carbo nate sequestratio n, (d) weat herability, (e) reverse weat heri ng, a nd (f) ratio of mari ne to terrestrial weat heri ng o n lo ng‐ter m stable state at mo- s p h eri c p C O 2 levels. Carbon input is the total ne w carbonfl ux added into the ocean ‐at mosphere syste m, and output is the total carbon fl ux re moved fro m the syste m ( w hic h eq uals i np ut at steady state).

ISS O N ET AL. 3 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

Fig ure 3. Co ntrols o n weat herability. For a set carbo n i nput, a ny i ncrease i n weat herability would lead to a drop i n stable state pC O 2 levels (a). Factors t hat have bee n proposed to reg ulate weat herability i ncl ude uplift rates (b), t he co mpositio n of t he weat herable cr ust (c), a nd t he degree of biological alteratio n (d). Hig her uplift rates a nd i ncreasi ng t he fractio n of exposed of basalts (s hifti ng ma fi c/felsic ratio) act to i ncrease weat herability. Hig her uplift rates lead to a hig her proportio n of fres h weat herable pri mary mi neral p hases (a nd a s horter soil reside nce ti me) wit hi n t he weat heri ng horizo n at Eart h's surface (e.g., Lee et al., 2015; Ma her & C ha mberlai n, 2014). I n pa nel (b), more i nte nse colors i ndicate a hig her fractio n of weat herable mi nerals wit hi n t he weat heri ng horizo n. Note t hat t he effects of uplift are more well establis hed t ha n i ncreasi ng basalt area.

T here has bee n a lo ngsta ndi ng debate about t he locus of silicate weat heri ng at Eart h's surface (e.g., Ku mp et al., 2000). T he vie w t hat silicate weat heri ng occurred predo mi na ntly i n terrestrial e nviro n me nts beca me the do minant vie w in the 1980s to 1990s, but recent work has stressed the i mportance of the weathering of silicate minerals in the marine environ ment ( Coogan & Gillis, 2018; Wall mann et al., 2008). Marine weath- ering can be categorized into t wo distinct marine environ ments and lithologies: (1) oceanic crust and (2) mari ne sedi me nts (sectio n 3.2.2).

2.2. Reverse Weathering T he for matio n of a ut hige nic silicate mi nerals co ns u mes alkali nity a nd ge nerates acidity, a process classically referred to as reverse weat heri ng ( Garrels, 1965; Macke nzie & Garrels, 1966a; Sillé n, 1961). As t he na me s ug- gests, t his process has, i n esse nce, t he opposite effect of silicate weat heri ng:

2 þ − ( 3) X þ 2 H 4 Si O 4 þ 6 H C O 3 ↔ X 3 Si 2 O 5 O Hð Þ4 þ 6 C O 2 þ 5 H 2 O

Clay aut hige nesis (equatio n (3) fro m left to rig ht) co nsu mes dissolved catio ns a nd co nverts carbo nate alka-

linity back into C O2 . T his C O2 no w has to facilitate another round of silicate weathering (equation (1)) before it can be re moved as carbonate rock. Reverse weathering in essence recycles carbon within the ocean‐at mosphere reservoir. This process makes carbon export fro m the syste m less efficient by increasing t he a mou nt of silicate weat heri ng required to sequester a n equal a mou nt of carbo n relative to a syste m wit h less exte nsive reverse weat heri ng (Isso n & Pla navsky, 2018). T h us, all else bei ng eq ual, a n i ncrease i n reverse

weathering will elevate the a mount of carbon in the ocean ‐at mosphere reservoir and thus baseline pC O 2

ISS O N ET AL. 4 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

Fig ure 4. Global carbon cycle box model de monstrating that carbon input must equal output (steady state) on geologic ti me scales (adapted fro m Berner & Caldeira, 1997). I n t he abse nce of a stabilizi ng feedback, s ustai ned i mbala nces (excess i np ut or o utp ut) will very rapidly lead to eit her r u na way ice ho use or greenhouse conditions. Color bar indicates % i mbalance bet ween input and output of carbon fro m the ocean‐at mosphere syste m. Blue indicates excess output and red excess i np ut.

levels ( Fig ures 1 a nd 2). It is i mporta nt to note, ho wever, t hat alt ho ug h reverse weat heri ng is a so urce of C O2 , it is not a source of ne w carbon and thus does not alter the total steady‐state fl ux of carbon entering (or leaving) the ocean‐at mosphere syste m. Reverse weathering does, ho wever, act as a net sink for ocean alkalinity. Si milar to silicate weathering, the process of reverse weathering enco mpasses an extensive suite of reactio ns— refl ecti ng t he diversity of clay mi neral species a nd fl exible mi neral stoic hio metries t hat can for m under different che mical conditions. Clays in the marine real m can for m directly fro m dissolved co nstitue nts or t hroug h catio n e nric h me nt of preexisti ng clays. 2.3. Mass Bala nce a nd Stabilizi ng Feedbacks T he mass of t he ocea n ‐at mosp here carbo n reservoir is s mall co mpared to t hat of geologic reservoirs, i mply- ing that short‐ter m mass i mbalances can easily occur. In a classic thought experi ment, Berner and Caldeira (1997) highlighted with a si mple box model that s mall carbon i mbalances will very rapidly (on <10 6 y e ars)

tri g g er C O2 ru na way i nto extre me ice house or hot house co nditio ns ( Figure 4). I n ot her words, it is not pos- sible for c he mical weat heri ng rates to be substa ntially out of bala nce wit h t he supply of C O 2 fro m volcanic and meta morphic sources for extended intervals without catastrophic consequences. While rapid te mpera- ture s hifts have bee n observed t hroug hout Eart h's history, t hese occurre nces are rare, a nd w here t hey have bee n observed to take place, Eart h's syste m has wit ho ut fail ure respo nded by reestablis hi ng relatively cle m- ent conditions. In other words, with li mited exceptions— fore most the Sno wball Earth Events ( Hoff man et al., 2017)— no ne of Eart h's cli mate pert urbatio ns were a t hreat to s ustai ned habitability. T his s uggests t hat although mass i mbalance may persist on the short ter m ( <10 6 years), the total a mount of carbon re moval

fro m the ocean‐at mosphere reservoir (F o ut ) must be nearly equal to ne w carbon input ( Fi n) i n t h e l o n g t er m ( > 1 06 years). It is t herefore reaso nable to ass u me t hat negative or stabilizi ng feedbacks m ust exist for a sys- te m to co nsiste ntly ac hieve mass bala nce. At t he core of t hese feedbacks is a li nk bet wee n t he rate‐li miti ng

st e p of C O2 dra wdo wn and C O 2 levels. T here are t hree stabilizi ng feedbacks t hat have bee n co m mo nly dis- c ussed ( Fig ure 5): 1. Terrestrial silicate weat heri ng fee dbac k . The vie w that the terrestrial silicate weathering feedback is t he mai n factor regulati ng te mperatures o n Eart h has bee n deeply e ntre nc hed i n our vie w of cli mate sta-

bility o n Eart h. T he basic idea of t his feedback is t hat a ny i ncreases i n C O 2 lead to i ncreased surface te m- peratures, acceleration of the hydrological cycle, and larger fl uxes of silicate che mical weathering,

ISS O N ET AL. 5 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

Fig ure 5. Sc he matic of t he processes t hat give rise to Eart h's t her mostat. T hree mai n stabilizi ng feedbacks associated wit h t he global carbo n cycle: terrestrial silicate weat heri ng (blue), mari ne silicate weat heri ng (red), a nd reverse weat heri ng (gree n).

leading to greater C O2 re moval fro m the ocean‐at mosphere syste m through continental weathering ( Ber ner et al., 1983; e.g., Urey, 1952; Wal ker et al., 1981). 2. Mari ne silicate weat heri ng feedbac k . Recent work has highlighted that rates of silicate mineral dis- sol utio n wit hi n mafi c ocea nic cr ust are se nsitive to botto m‐water te mperat ures, esse ntially establis hi ng a

negative feedback with C O 2 give n t hat sea surface a nd botto m water te mperatures are i nti mately coupled ( Brady & Gíslason, 1997; e.g., Coogan & Dosso, 2015). Silicate dissolution within marine sedi ments has also been noted to contribute to global alkalinity fl uxes (Solo mon et al., 2008; Wall mann et al., 2008).

Here, release of C O 2 during organic matter deco mposition drives silicate mineral dissolution (e.g., Aloisi et al., 2004). Additio nally, t he prod uctio n of dissolved orga nic h u mic a nd f ulvic acid a nio ns t hat c o m pl e x c ati o ns ( e. g., Al 3 + ) is proposed to enhance mineral dissolution through further undersaturation of pri mary silicate p hases. Rates of pri mary productivity (a nd t he myriad factors affecti ng water colu m n organic carbon re mineralization) have therefore been proposed to exert a control on weathering rates ( Wall mann et al., 2008). There are links bet ween organic matter delivery to sedi ments and te mpera- ture— ho wever, these associations have not yet been de monstrated to clearly give rise to a stabilizing feedback. Direct botto m‐water te mperature a nd p H co ntrols o n mi neral dissolutio n rates oug ht to estab-

lish a negative feedback with at mospheric C O2 levels as well, b ut t he c he mistry of pore waters (a nd waters wit hi n fract ured ocea nic cr ust) ca n evolve sig ni fi ca ntly wit h dept h, a nd do not necessarily refl ect s urface water co nditio ns.

3. Reverse silicate weatheri ng feedback . Theflux of C O 2 to the ocean‐at mosphere syste m derived fro m reverse weathering is sensitive to marine p H conditions and thus pC O 2 (Isso n & Pla navsky, 2018; Sillé n, 1961, 1967). T his li nk res ults i n a stabilizi ng cli mate feedback. For i nsta nce, a n i ncrease i n pC O 2 l e v els (decrease in marine p H) will lead to decreased reverse weathering, in turn acting to lo wer pC O 2 l e v els. Beca use i ncreasi ng te mperat ures will act to elevate rates of reverse weat heri ng (reactio ns are more rapid at hig her te mperatures), t he p H ki netic effect must out weig h t he te mperature effect. Reverse weat heri ng has been de monstrated to be sensitive to p H in the most detailed clay experi mental work done thus far (e.g., Tosca et al., 2011; Tosca et al., 2016), t ho ug h it s ho uld be noted t hat t here is a deart h of clay ki netic d at a.

ISS O N ET AL. 6 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

T a bl e 2 Modern Global Carbon Cycle Budget Source/sink Fl u x ( 1 0 1 2 mol/year) Reference

CO 2 s o ur c es Collisional meta morphis m 0.5–7.0 Becker et al. (2008) and Ste wart and Ague (2018) Subduction meta morphis m 0. 3 – ≫ 1. 0 Kele men and Manning (2015) ( diff us e) Arc volcanis m 1. 5 – 3. 5 Dasgupta and Hirsch mann (2010), Hilton et al. (2002), and Marty and Tolstikhin (1998) Mid ‐ocean ridge 1. 0 – 5. 0 Chavrit et al. (2014), Dasgupta and Hirsch mann (2006, 2010), Hirsch mann (2018), Marty and Tolstik hi n (1998), Matt he ws et al. (2017), a nd Saal et al. (2002) Ocean Island 0. 1 2 – 3. 0 Dasgupta and Hirsch mann (2010) and Marty and Tolstikhin (1998) Reverse weathering 0. 5 – 1 0. 0 Rah man et al. (2016) and Tréguer and De La Rocha (2013) ( Alk:Si = 1–2) − HCO 3 s o ur c es Terrestrial weat heri ng 1 1. 5 – 2 3 Gaillardet et al. (1999) a nd Li a nd Elder fi eld (2013) (not including ground water) Oceanic crust weathering 0.2 –3.75 Coogan and Gillis (2018) Marine sedi ment weathering 5 –20 Wall mann and Aloisi (2012) and Wall mann et al. (2008) − HCO 3 si n ks Biogenic carbonate 1 4 – 2 5 Wall mann and Aloisi (2012) Authigenic carbonate 0. 5 – 1. 5 Sun and Turchyn (2014) (s e di m e nt) Authigenic carbonate 1. 5 – 2. 4 Coogan and Gillis (2018) (ocea nic crust)

All t hree of t hese stabilizi ng feedbacks operate toget her as a u ni fi ed global silicate weat heri ng feedback, alt houg h wit h varied respo nse ti mes for eac h i ndividual feedback. Wit h a s hift a way fro m a stable state i n

the carbon cycle and cli mate (e.g., with a change in C O2 outgassing rates) the change in at mospheric p C O 2 is set by t he stre ngt h of t he global silicate weat heri ng feedback. A stro nger silicate weat heri ng feed- back yields a s maller change in pC O 2 level for a n eq uivale nt forci ng. Co ntrols o n t he stre ngt h of t he silicate weat heri ng feedback traditio nally i nclude t he a mou nt of fres h silicate material exposed at Eart h's surface u ndergoi ng water ‐rock i nteractio ns, t he co mpositio n of Eart h's weat herable s hell, a nd t he role of t he terres- trial biosp here i n mediati ng weat heri ng rates (see Ku mp et al., 2000). Belo w we develop t he idea t hat mari ne che mistry— and in particular marine Si concentrations— will also play a role in controlling the strength of t he global silicate weat heri ng feedback.

3. Global Carbo n Cycle Budget 3.1. Carbo n Sources 3.1.1. Modern Degassing Esti mates

CO 2 is released fro m the solid Earth into surface reservoirs via both volcanic and meta morphic processes (equatio n (1), fro m rig ht to left; Table 2). Volca nic fl uxes have lo ng bee n co nsidered t he pri mary co ntribu- tio n to global degassi ng, a nd, as suc h, t here is a substa ntial body of work dedicated to co nstrai ni ng t heir

magnitudes. Modern mid ‐ocean ridge volcanis m is esti mated to contribute ~1.0 to ~5.0 T mol C O 2 / y e ar (for reference 1 T mol C is approxi mately 0.012 Pg C) ( Chavrit et al., 2014; Dasgupta & Hirsch mann, 2006, 2010; Hirsch mann, 2018; Le Voyer et al., 2017; Marty & Tolstikhin, 1998; Matthe ws et al., 2017; Saal et al., 2002). W hile t his ra nge is q uite sig nifi ca nt, it is possible t hat t his upper li mit re mai ns co nservative, give n t he lac k of co nsideratio n of ot her factors s uc h as b ubble loss (e.g., C havrit et al., 2014). F urt her, explo-

sive volcanis m along ridges provide evidence for melt inclusions with extre mely high C O 2 co nte nt, s uggest- ing that the C O2 content of mid‐ocean ridge basalt may be much more heterogeneous than previously t ho ug ht ( Helo et al., 2011; Ilyi nskaya et al., 2018).

Arc volcanic degassing esti mates range fro m ~1.5 to ~3.2 T mol C O 2 /year ( Dasgupta & Hirsch mann, 2010; Hilton et al., 2002; Marty & Tolstikhin, 1998). Ho wever, there is work suggesting that arc mag mas may be

m or e C O 2 rich than previously hought, and that the arc volcanic C O 2 flux may be correspondingly higher ( Blundy et al., 2010). A large portion of the carbon e mitted by these volcanoes may be derived fro m

ISS O N ET AL. 7 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

re worki ng of crustal carbo nate rocks i n t he upper plate ( Maso n et al., 2017). Furt her, meta morp his m i n co n- tact aureoles around mag matic intrusions also drives decarbonation reactions ( D' Errico et al., 2012; Lee et al., 2013). We are not a ware of a ny q ua ntitative esti mate of t he global co ntact meta morp hic fl ux ( fl ux esti-

mates fro m observations of at mospheric plu mes above volcanoes should integrate this C O 2 with mag matic CO 2 ). For exa mple, Nesbitt et al. (1995) esti mate t hat 1.5 to 7.8 T mol C O2 /year were released by Cenozoic contact meta morphis m in the North A merican Cordillera. Lee et al. (2013) suggest that contact meta morph-

is m in continental arcs has been a major control on at mospheric C O2 throughout the Phanerozoic. Intraplate volcanis m is less well studied, but Dasgupta and Hirsch mann (2010) suggest a range of ~0.12 to

~3.0 T mol C O 2 /year degassed fro m ocean island basalts and Lee et al. (2013) calculate a fl ux of ~0.8 to ~2.4 T mol C O 2 /year fro m co nti ne ntal rift volca nis m. Meta morphic decarbonation in mixed carbonate ‐silicate rocks has been sho wn to be an equally i mportant

source of C O2 ( Bickle, 1996; Kerrick & Caldeira, 1993, 1999; Kerrick & Caldeira, 1998; Ste wart et al., 2019). While it has been suggested that meta morphic decarbonation is negligible except at very high te m- perat ures (e.g., Dasg upta, 2013), t hese calc ulatio ns ig nore t he critical effect of fl uid i n fi ltratio n. A rock t hat is in contact with a water‐bearing fl uid during meta morphis m in an open syste m will begin to decarbonate at

much lo wer te mperatures (Ferry, 1988) and may release more than 500 % more C O 2 than meta morphis m of t he sa me rock i n a closed syste m (Ste wart & Ague, 2018). T he exte nt to w hic h t he Urey reactio ns ca n work i n reverse depe nd o n t he supply of Si eit her fro m i mpurities i n t he rock or tra nsported i n a fl uid ( Ague, 2003). As noted by Ste wart and Ague (2018), multiple independent calculations of degassing during regional meta- 6 6 − 1 − 2 morphis m agree on an area ‐nor malized flux of ~0.5 × 10 t o ~ 7 × 1 0 m ol C O 2 · y e ar · k m ( Bec ker et al., 2008; C hiodi ni et al., 2000; Kerrick & Caldeira, 1998; Skelto n, 2011). T hese st udies esti mate t he fl ux via bot h ther modyna mic modeling of meta morphis m at depth ( Kerrick & Caldeira, 1998; Skelton, 2011; Ste wart & Ague, 2018) and surface measure ments of degassing in modern mountain belts (Becker et al., 2008;

C hiodi ni et al., 2000), s uggesti ng t hat most of t he devolatilized C O 2 does escape to t he surface e nviro n me nt. Multiplying this areal fl ux by the area of present ‐day orogenesis (~10 6 k m 2 , do minated by the Hi malayas)

( Becker et al., 2008), we arrive at a modern regional meta morphic fl ux of 0.5 to 7.0 T mol C O 2 / y e ar. Carbon dioxide is also released via meta morphic decarbonation in subducting oceanic crust. Based on closed‐syste m ther modyna mic modeling, it was believed that this decarbonation was mini mal and that most

CO 2 is retained within the subducting slab to depths of more than ~100 k m ( Dasgupta & Hirsch mann, 2010; e.g., Kerrick & Connolly, 2001). Ho wever, recent work by Kele men and Manning (2015) suggests that sub- d uctio n zo nes are hig hly i neffi cie nt at s h uttli ng carbo n i nto t he co nvecti ng ma ntle. T here are several possi- ble explanations for this discrepancy. As with regional meta morphis m, subduction‐related decarbonation may also be facilitated by fl uid infiltration ( Gor man et al., 2006), thus accounting for the presence of a water ‐bearing fl uid increases fl ux esti mates. In addition, congruent dissolution of carbonate minerals during

subduction may release even more C O 2 fro m the slab ( Ague & Nicolescu, 2014). So me a mount of C O2 is als o released during partial melting of carbonate‐bearing rocks (e.g., Duncan & Dasgupta, 2014; Grassi & Sch midt, 2011; Poli, 2015). Further more, mechanical processes such as sedi ment diapiris m may transfer

CO 2 fro m the subducting slab to the overriding plate ( Kele men et al., 2003). In an apparent contradiction to the“what goes do wn mostly co mes back up ” world vie w ( Kele men & Manning, 2015), global geoche mical

argu ments (specifically relating to C O 2 / Ba ratios) may indicate that 35–80 % of outgassed C O 2 m ust b e returned to the mantle, presu mably via subduction ( Hirsch mann, 2018). Moreover, C O 2 released via decar- bonation and dissolution may be reincorporated into the slab by carbonation reactions driven by migrating

fl ui ds (e.g., Piccoli et al., 2016; Sca mbell uri et al., 2016). Fi nally, t he a mo u nt of C O 2 a nd orga nic carbo n deliv- ered to subductio n zo nes varies co nsiderably i n bot h space a nd ti me (Pla nk & Ma n ni ng, 2019). T he storage of

CO 2 i n t he lit hosp here is a major u ncertai nty i n bot h models. T h us, t he effi cie ncy of C O 2 subduction re mains an area of ongoing research and debate, and more work is needed to reconcile these disparate models.

Once released, C O 2 fro m t he slab may escape to t he surface as part of t he arc volca nic fl ux or it may for m a separate, diffuse C O 2 flux. Kele men and Manning (2015) esti mate this diffuse subduction flux as ~0.3 to ~1.0 T mol C O 2 /year but go o n to speculate t hat it may be muc h hig her. It is also probable t hat so me slab‐ d eri v e d C O 2 is precipitated in the overlying crust for long‐ter m storage (up to ~3.9 T mol C O2 / y e ar fr o m Kele men & Manning, 2015), rather than being released to the surface. The proportion of C O 2 t hat follo ws each of these paths re mains as a major uncertainty in our understanding of global carbon cycling.

ISS O N ET AL. 8 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

3.2. Carbo n Si nks 3.2.1. Terrestrial Weat heri ng As o utli ned above, t he weat heri ng of silicate mi nerals i n terrestrial setti ngs co ntrib utes al kali nity to t he mar- ine environ ment ( Table 2). Carbonate weathering coupled to the eventual precipitation of carbonate rock fro m sea water, ho wever, has no long‐ter m net effect as a source or sink of carbon unless these fl uxes are out of balance— i mplying a change to the co mposition of Earth's sedi mentary reservoir. Further, changes to the co mposition of the upper continental crust have been suggested to infl uence the weatherability of

Earth's surface and hence at mospheric pC O 2 ( Gaillardet et al., 1999; Hart ma n n et al., 2014; e.g., S uc het et al., 2003). Today, the upper continental crust exposes ~65–71 % sedi mentary rock ( Hart mann et al., 2014; Hart mann & Moosdorf, 2012; Li, 2000; Suchet et al., 2003) (Figure 6). The nonsedi mentary fraction co mprises ~16 % basic (e.g., basalts a nd gabbros), ~8 % i nter mediate, a nd 26 % acidic (e.g., r hyolites a nd gra n- ites) igneous rocks, with meta morphic rocks making up the re maining ~50 % ( Hart mann et al., 2014). On average, t he mai n co mpo ne nts of gra nite are K‐feldspar, sodic plagioclase a nd quartz (i n roug hly equal pro- portio ns), w hile basalts are hig h i n calcic plagioclase (~60 %), pyroxe ne (~30 %), a nd olivi ne (~10 %) ( Maso n & Moore, 1982). The sedi mentary reservoir co mprises roughly 51 wt. % shale, 23 wt. % sandstone, 25 wt. % car- bonate, and 1 % evaporate ( Li, 2000). It should be noted that the meta morphic rock classi fi cation is co mpo- sitio nally broad (i.e., ca n refer to a rock of virt ually a ny co mpositio n) a nd basically describes a ny rock wit h a crystalli ne texture i nterpreted to have bee n subject to elevated pressure a nd/or te mperature so meti me after initial for mation of the rock ( Hart mann & Moosdorf, 2012). Meta morphic rocks are ~90 % metasedi mentary, wit h metaig neous rocks co mprisi ng t he re mai ni ng ~10 % ( Ro nov et al., 1990), w hic h bolsters t he case t hat t he Earth's weatherable shell is do minated by sedi mentary (and metasedi mentary) rocks (e.g., Bluth & Ku mp, 1991; Bluth & Ku mp, 1994).

In the modern oceans, calciu m ( Ca) is the main cation involved in the re moval of carbon as carbonate. S hales a nd silts are, ho wever, o n average, depleted i n Ca relative to crystalli ne rocks. T his does not mea n t hat t he weat heri ng of s hales a nd silts does not co ntribute to t he for matio n of carbo nate rock. T he weat heri ng of s hales a nd silts has to be li nked to eit her: (1) t he for matio n of dolo mite ( Ca‐Mg carbo nate), w hic h, alt ho ug h unco m mon today, was far more pro minent during other intervals in Earth's history when dolo mite was the do minant carbonate phase (Lo wenstein et al., 2003) and/or (2) hydrother mal exchange of Mg (derived fro m s hales or silt weat heri ng) for Ca eve ntually sequestered as calciu m carbo nate.

Tre mendous effort has been put into understanding the factors controlling terrestrial silicate weathering rates (e.g., Bra ntley, 2008; K u mp et al., 2000; Li et al., 2017). Arg u me nts ce nter o n t he relative i mporta nce of soil bio me, te mperature, denudation rates, tectonics, and hydrologic controls on che mical weathering: T his debate is nua nced, eve n t houg h t he basic idea of ho w eac h o ne i n fl ue nces weat heri ng is straig htfor ward. Mi neral dissolutio n rates i ncrease wit h te mperature, orga nic acid levels, de nudatio n, a nd i ncreased water flux through the weathering zone (e.g., Brantley, 2008; Chen & Brantley, 1997; Maher & Cha mberlain, 2014). O n a broad scale, if de n udatio n rates are hig her, t here is more fres h material to weat her, a nd t h us, sili- cate weat heri ng rates are likely to i ncrease. Si milarly, if more water is fl uxed t hroug h t he weat heri ng zo ne, mi nerals will be furt her fro m equilibriu m a nd dissolutio n rates will i ncrease ( Ma her & C ha mberlai n, 2014). 3.2.2. Tra nsport Versus Weatheri ng Li mitatio n There are t wo end me mber vie ws for thinking about controls on che mical silicate weathering in terrestrial environ ments— supply‐li mited and weathering‐li mited regi mes ( Ku mp et al., 2000; Stallard, 1985; Stallard &

Ed mo nd, 1981, 1983). I n a supply ‐li mited syste m, rates of C O2 dra wdo wn are li mited by the a mount of mate- rial exposed to weathering, and changes to at mospheric pC O 2 levels (a nd surface te mperature) do not i nfl u- e nce weat heri ng rates. I n t his e nd me mber vie w, it is t he exte nt of tecto nic uplift a nd he nce t he supply of

fresh material that regulates at mospheric pC O 2 levels— esse ntially negati ng t he terrestrial silicate weat her- ing feedback. In contrast, a weathering‐li mited regi me has an“excess” of material available for water‐rock

interactions, and rates of C O2 consu med through silicate weathering are li mited by both te mperature and the supply of C O2 (co upled to factors s uc h soil p H a nd mi neralogy). I n t his world vie w, t here is a n active ter- restrial silicate weathering feedback, where an increase in at mospheric C O 2 levels and global te mperature will i ncrease C O 2 delivery to and consu mption within the weathering real m. T here is co mpelli ng evide nce to suggest t hat global silicate weat heri ng today is not supply ‐li mited. Global data sets i ndicate t hat a n i ncrease i n r u noff is met by a n i ncrease i n silicate weat heri ng rates (e.g., Ma her

ISS O N ET AL. 9 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

Fig ure 6. Co mposition of the modern upper continental crust (Suchet et al., 2003). Rock exposure at Earth's surface is do minated by sedi mentary rock. This includes the meta morphic co mponent, which is esti mated to consist of ~90 % metasedi mentary rock.

& Cha mberlain, 2014), i mplying that more intense hydrology al ways increases C O 2 ne utralizatio n, a nd so Earth as a whole cannot be supply ‐li mited (Figure 7). Quantifying the extent of physically eroded but che mically un weathered material accu mulating in river deltas can be used to assess the extent of local supply li mitation on che mical weathering rates. In addition, the expression of i m mature soils rich in pri mary silicate p hases at Eart h's surface, particularly i n hig hla nd areas, represe nt a n excess of silicate material available for weathering. Further more, geoche mical and mineralogical evidence suggest that s hifts i n global te mperatures across hypert her mal eve nts are ofte n coupled to accelerated rates of silicate weathering (e.g., Pen man, 2016; Von Strand mann et al., 2013). Eart h's syste m as a w hole is ge nerally vie wed to sit so me w here i n bet wee n these t wo end me mber regi mes, with so me environ ments sitting closer to one end me mber than another denoting local scale heterogeneity to Earth's weatherable environ ment. The A mazon watershed, for instance, sits closer to t he supply‐li mited e nd me mber regi me relative to t he Great Hi malayan Watershed ( Hart mann et al., 2014). Such supply ‐li mited envir- on ments are characterized by thicker, more mature (che mically depleted) soils largely devoid of pri mary silicate mi nerals (e.g., feldspars, olivi ne, and hornblende), and are co mposed mostly of secondary mineral phases (e.g., clay mi nerals a nd oxides) depleted i n soluble catio ns ( Ca2 + , M g2 + , N a + , a n d K+ ). I n co ntrast, waters heds t hat sit closer to t he weat heri ng li mited environ ment have a less mature, higher pri mary silicate mineral co nte nt (less c he mically depleted).

T he globally averaged weat heri ng state has t he pote ntial to slide bet wee n t hese t wo e nd me mber regi mes. I n partic ular, t here are i ntervals i n Eart h's history where Earth as a whole is suggested to have shifted to ward the supply li mited end me mber regi me. These include the (1) end‐Per mian

to early Triassic interval where the a mount of C O2 injected into the ocean‐at mosphere syste m is suggested to have co me close to “exhausting” Fig ure 7. Global geoc he mical river data sets i ndicate t hat Eart h's syste m as Eart h's weat herable s hell so as to provide a n expla natio n for t he delayed a w hole is not s upply li mited. I n a s upply li mited Eart h syste m, a n i ncrease cli mate recovery during this interval ( Ku mp, 2018) and (2) end‐ i n rai nfall ( more i nte nse hydrology) will not be met by a n i ncrease i n c he- mical weat heri ng rates. Eart h's river syste m ho wever, i ndicates a stro ng Cretaceous through early Eocene ti me periods ( Misra & Froelich, 2012). correlatio n bet wee n t he acceleratio n of t he hydrological cycle a nd c he mical It is perhaps i mportant to note that although Earth's syste m as a whole weathering rates (data fro m Maher & Cha mberlain, 2014). is not supply li mited, this does not mean that Earth's cli mate syste m is

ISS O N ET AL. 1 0 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

i nse nsitive to tecto nic c ha nges. All else bei ng eq ual, a n i ncrease i n tecto nic uplift rates will elevate bot h weathering fl uxes globally, and the weatherability of Earth's surface environ ment and thus the strength of the silicate weathering feedback (e.g., Caves et al., 2016; Ku mp & Arthur, 1997; Rugenstein et al., 2019; Zhang & Planavsky, 2019b). Rates of uplift bring fresh silicate material into the weatherable portion of Eart h's s urface e nviro n me nt. I n esse nce, a n i ncrease i n uplift will decrease t he i ntegrated mat urity of soil profi les globally. 3.2.3. Terrestrial So urces of Acid for Weat heri ng

Co nti ne ntal weat heri ng reactio ns are tied to t he ne utralizatio n of C O 2 . Carbon dioxide is supplied predo mi- nantly through direct diffusion bet ween the at mosphere and terrestrial fl uids, the partitioning of C O 2 i nt o rai n water, a nd/or t he s upply a nd oxidatio n of orga nic matter i n soils (e.g., t hro ug h respiratio n) ( Fig ure 8).

A n i ncrease i n t he biological supply of C O 2 to t he soil e nviro n me nt has t he pote ntial to alter t he p H of soil pore water a nd t hus mi neral dissolutio n rates. O n t his basis, t here have bee n argu me nts co ncer ni ng t he role t hat biology ( microbiota a nd vascular pla nts) has played i n alteri ng bot h weat heri ng rates a nd t he weat her-

ability of Earth's terrestrial environ ment and in turn at mospheric pC O 2 ( a n d O2 ) l e v els. Perhaps most notably, it has bee n co m monly proposed t hat the e merge nce and diversi fi catio n of vascular land

plants during t he Siluria n and Devonia n (~420 millio n years ago) could have led to a dra wdo wn of pC O 2 l e v els t hro ug h a tra nsie nt e n ha nce me nt of terrestrial weat heri ng rates ( Algeo et al., 1995; Ber ner, 1997, 1998). It is s ug- gested that this transient weathering enhance ment could occur through several mechanis ms including (1) i ncreasing water retentio n i n soils, (2) root a nd heterotrophic respiratio n of pla nt debris, (3) productio n of orga nic acids fro m pla nt roots, (4) root p hysical erosio n, a nd (5) stabilizatio n of soils ( Ber ner, 1992; Drever, 1994; Gibling & Davies, 2012; Griffi t hs et al., 1994; Wi n nick & Ma her, 2018). S upport for t his argu me nt typically i nvolve calli ng upo n bot h laboratory (e.g., Q uirk et al., 2015) a ndfi eld data (e.g., Bor ma n n et al., 1998; Moulto n & Ber ner, 1998) t hat i ndicate e nhanced che mical weathering i n syste ms wit h land plants as co mpared to barre n surfaces or o nes covered wit h moss/lic he n (i.e., no sig nifi cant root pe netratio n) ( Le nton, 2001). I n contrast, rece nt steady‐state carbon cycle modeli ng work i ndicate t hat t he rise of land plants pote ntially decreased

(i nstead of i ncrease) silicate weat heri ng rates, a nd t hat a ny drop i n C O2 levels was i nstead li nked to a n i ncrease i n t he stre ngt h of t he silicate weat heri ng feedback ( D' A nto nio et al., 2019).

The link bet ween vascular plants and at mospheric C O 2 levels through the Paleozoic has, ho wever, re mained contentious for several reasons. First, the earlier evolving bryophytes (liver worts; in the Ordovician) ( Quirk et al., 2015) have also been sho wn to accelerate weathering rates, and direct co mpar- ison of these with the effect of vascular plants ( with more deeply penetrating roots) has proven challen- ging. Further, integrating these biologically enhanced weathering rates derived fro m short lived experi ments (“instantaneous” relative to the ti me scale of soil for mation) into long‐ter m global biogeo- che mical cycle models that operate on the multi million‐year ti me scales has the potential to drastically overesti mate the integrated weathering rate. Silicate weathering rates decrease as a soil horizon matures given that pri mary phases beco me more depleted (Figure 9). In areas with li mited denudation, integrated or lo ng‐ter m weat heri ng rates wit h a nd wit hout la nd pla nts will be closer t ha n t he observed i nitial differ- ence. Although both field and experi mental data that highlight biotic enhance ment of weathering (so me- ti mes by several orders magnitude), rate enhance ments at such high levels are likely not sustainable on longer ti me scales. When weathering fluxes are integrated on ti me scales si milar to or longer than that of soil for matio n (a nd maturatio n), t he net weat heri ng effect wit h or wit hout vascular pla nts beco mes sig- nifi cantly less or even indistinguishable (Figure 9). This fra me work was put into question by Keller and Wood (1993), where they used a reactive transport model to make a case that vascular plants are likely not required to mai ntai n lo w soil p H. I nstead, t hey propose t hat terrestrial c he mical weat heri ng rates were more likely eq ually i nte nse before a nd after t he rise of vasc ular pla nts. Speci fi cally, t hey posit t hat beca use

CO 2 loss fro m soils is a slo w process, o nly a s mall a mo u nt of microbial soil respiratio n is req uired to mai n- tain high C O2 (lo w p H) pore waters. On the e mpirical side, terrestrial cyanobacterial biofil ms have been fo u nd to sig nifi ca ntly i ncrease silicate weat heri ng rates (Seiffert et al., 2014).

S ulf uric acid ( H 2 SO 4 ) derived fro m t he oxidatio n of pyrite:

F e S 2 þ 3 :7 5 O 2 þ 3 :5 H 2 O ↔ 2 H 2 SO 4 þ F e ð O H Þ3 ( 4)

also accou nts for a portio n of t he acid t hat facilitates weat heri ng i n t he terrestrial e nviro n me nt (~5 % of

ISS O N ET AL. 1 1 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

Fig ure 8. Sources of acidity for terrestrial silicate weat heri ng: co ntributio ns fro m rai n a nd soil respiratio n. Results fro m a si mple water rock mass bala nce model explori ng t he a mou nt of rai n required to ge nerate a global terrestrial weat heri ng alkali nity fl ux of 10 T mol/year t hroug h terrestrial silicate weat heri ng. Ne w C O 2 i nput i nto t he syste m is set at 10 T mol/year, a nd all i nfi ltrati ng C O2 is assu med to be neutralized. We specify that 50 % of C O2 neutralization occurs through the weathering of carbonate rock, meaning that 20 T mol/year of C O 2 has to be delivered to t he terrestrial e nviro n me nt. Red circle i ndicates prei nd ustrial val ues of rainfall = 1 m/year, te mperature = 12 ° C, and at mospheric pC O 2 = 280 pp m. This model assu mes quantitative neutralization of rain water through weathering. The model outputs indicate that at pC O 2 levels of 280 pp m, 10 m/year of rai nfall is req uired to ne utralize volca nic C O2 if here was no soil respiratio n. T his is ro ug hly a n order of mag nit ude hig her t ha n moder n average a n n ual rai nfall esti mates. To ac hieve a n a n n ual rai nfall esti mate of ~1 m/year, t his req uires t hat rai n s upplies o nly 3 T mol/year of C O 2 for silicate weathering (right panel). The re maining 7 T mol/year of C O2 ca n be accou nted for by acid ge neratio n i n terrestrial pore waters t hroug h soil respiratio n. T his si mple model i ndicates t hat rai nfall is not t he do mi na nt source of C O2 for weat heri ng today (red circle does not sit o n t he red li ne). Ho wever, at high pC O 2 conditions the a mount of rainfall needed to deliver enough C O 2 to t he critical zo ne to e ns ure a bala nced carbo n cycle (silicate weat heri ng equal to C O 2 o utgassi ng) is sig nifi ca ntly less (left pa nels). U nder t hese co nditio ns, rai nfall beco mes t he do mi na nt so urce of C O 2 , a n d c h a n g es i n s oil r es pir ati o n will play a s maller role. The gray region indicates possible te mperature and C O 2 co mbi natio ns assu mi ng differe nt cli mate se nsitivities (2–4.5 ° C) at moder n solar lu minosity.

ISS O N ET AL. 1 2 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

Fig ure 9. (a, b) The effect of land plants on weathering during soil maturation. Laboratory and fi eld work both indicate that vascular plants accelerate che mical weat heri ng of bedrock over t he s hort ti me scales i n w hic h t hese experi me nts are co nducted ( <1 year). It is i mporta nt to note t hat t he observed weat heri ng rates are not likely s ustai nable over t he ti me scale of soil for matio n (o n a ti me scale of typically t ho usa nds of years). As soil pro fi les mat ure, t hey beco me more c he mically depleted (i n alkali nity a nd i n fres h material) a nd t herefore t he rate of c he mical weat heri ng will be da mpe ned. (c) Results fro m a reactio n tra nsport model s ho wi ng t he relatio ns hip bet wee n t he material age or soil reside nce ti me ( horizo ntal axis) a nd fres h mi neral co nce ntratio n ( Xs/ Xr) ( Ma her & C ha mberlai n, 2014). T he upper a nd lo wer bracketi ng li nes represe nt se nsitivity a nalysis to t he weat heri ng rate co nsta nt (K eff ) of K eff / 2 a n d K eff × 2 fro m t he best fi t val ue, respectively, wit h all model ru n results falli ng bet wee n t hese upper a nd lo wer brackets.

silicate dissolutio n a nd 10 % of carbo nate dissolutio n) ( Bra ntley, 2008; Torres et al., 2014). Iro n oxidatio n likely accounts for a s maller portion of the acid that facilitates weathering, although this fl ux is poorly constrained. 3.2.4. Grou nd water Alkali nity Fluxes There have been extensive studies of river input to the ocean and its in fl uence on global biogeoche mical cycling (e.g., Berner & Berner, 2012; Gaillardet et al., 1999; Meybeck, 1987; Milli man & Farns worth, 2011), but we still have a relatively nascent understanding of ground water discharge and ele ment fluxes to the ocean ( Taniguchi et al., 2002). Ho wever, there is mounting evidence that ground water discharge is a critical co mpo ne nt of t he water a nd nutrie nt budget i n t he la nd‐ocea n syste m (e.g., Bur nett et al., 2003; K wo n et al., 2014; Moore et al., 2008). Furt her, t here is evide nce t hat grou nd water plays a sig ni fi ca nt role in coastal ecosyste m evolution ( Moosdorf & Oehler, 2017; Paytan et al., 2006). Ground water contains, on average, significantly higher bicarbonate concentrations than overlying surface waters, indicating that weathering occurs not only in soil horizons but also deeper within the crust ( Zhang & Planavsky, 2019a). I n ter ms of t he global carbo n cycle, co nti ne ntal gro u nd water disc harge serves as a pat h way for dissolved car- bon fl uxes to the ocean where carbonate eventually precipitates, and may therefore likely provide a signi fi - cant but previously underappreciated carbon sink ( Zhang & Planavsky, 2019a).

Ground water discharge to the ocean is usually called sub marine ground water discharge (S G D), which is defined as “any and all flo w of water on continental margins fro m the seabed to the coastal ocean” ( B ur nett et al., 2003). Si nce gro u nd water fl o w i n a coastal regio n ca n be drive n by bot h terrestrial a nd mari ne forces, S G D t herefore co mprises t wo sources— terrestrially derived fres h water (fres h S G D) t hat is origi nally produced by rain water infiltration on land, and modifi ed sea water (saline S G D) that penetrates through the

ISS O N ET AL. 1 3 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

la nd‐ocea n i nterface a nd is recirculated back to t he ocea n. W he n t hese t wo sources are mixed, S G D is called mixed or brackish S G D. There are several approaches to esti mate S G D, depending on the spatial extent of i nterest. O n a local scale, seepage meters ca n meas ure mixed S G D at specifi c locatio ns directly. O n a regio nal scale, c he mical tracers (e.g., radiu m a nd rado n) ca n be used to qua ntify t he mixed S G D fl uxes i ntegrated over continental shelves. On a global scale, hydrological modeling and water balance methods can both be applied for esti mation of fresh S G D ( Mulligan & Charette, 2009; Taniguchi, 2002). Unfortunately, flux esti mates fro m these approaches are typically significantly different (see discussion in Zhang & Planavsky, 2019b) Co mpared with river syste ms that are well gauged and monitored throughout the world, direct quanti fica- tio n of global S G D is extre mely diffi cult si nce it mai nly occurs as u nsee n diffusive fl o w o n co nti ne ntal mar- gi ns a nd varies sig nifi ca ntly over te mporal a nd spatial scales. T he i nstallatio n of seepage meters lacks global coverage, and upscaling of seepage measure ment re mains a challenge. The che mical tracer approach see ms to be more fr uitf ul i n q ua ntifyi ng large‐scale S G D. Usi ng 2 2 8 Ra as a tracer, Moore et al. (2008) co ncl uded t hat the magnitude of S G D in coastal region of the Atlantic Ocean is co mparable to the corresponding river fl ux. Co mbining global observations of 2 2 8 Ra a nd a n i nverse model, K wo n et al. (2014) argued t hat t he S G D alo ng t he coastli nes bet wee n 60°S a nd 70° N is aro u nd 3 ti mes greater t ha n t he river disc harge. After correcti ng t he sali nity effect, C ho a nd Ki m (2016) stated t hat i nstead of 3 to 4 ti mes, global S G D is approxi mately 1–1.5 ti mes t he river disc harge. I n ter ms of ele me nt fl uxes, it is reported by C ho et al. (2018) t hat dissolved i nor- ganic nitrogen, phosphorus, and silicon derived fro m S G D are co mparable with the river fluxes to the global ocea n. For dissolved inorganic carbon ( DI C) and alkalinity, several studies argued that S G D ‐derived fl uxes exceed regio nal river i np ut a nd i n so me cases are 1 order of mag nit ude larger ( Li u et al., 2014; Ste wart et al., 2015). It s hould be noted t hat a large portio n of S G D is t houg ht to belo ng to sali ne S G D ( K wo n et al., 2014; Ta niguc hi et al., 2006). Co nseq ue ntly, fres h S G D mig ht be sig nifi ca ntly less t ha n river disc harge ( Ta nig uc hi, 2002), a nd the ele mentfl ux (including dissolved carbon) derived fro m fresh S G D might be di minished. Given the fact t hat DI C co nce ntratio n i n gro u nd water is typically hig her t ha n rivers i n t he sa me regio n (Ste wart et al., 2015; Szy mczycha & Pe mpko wiak, 2015), even a relatively s mall magnitude of fresh S G D could potentially make a nonnegligible contribution to the total dissolved carbon fl ux to the ocean. It was recently esti mated that bet ween 20 % and 250 % of che mical silicate weathering globally occurred in ground waters (Zhang & Planavsky, 2019a). This esti mate highlights fore most that large uncertainties exist for esti mating fresh S G D a nd t he exte nt of grou nd water mediated silicate weat heri ng. Ho wever, it has beco me clear over t he past decade t hat t his fl ux is likely a sig ni fi ca nt part of t he global carbo n cycle. 3.2.5. Weathering of Marine Sedi ments The e merging perspective fro m recent studies is that silicate mineral dissolution within marine environ- me nts co ntrib utes to global alkali nity a nd t h us carbo n seq uestratio n ( Fig ure 1), s uggesti ng t hat t he terres- trial to mari ne tra nsitio n is better vie wed as a co nti nuu m rat her t ha n t wo disti nct e nviro n me nts. T he p hysical erosio n of roc ks i n terrestrial e nviro n me nts delivers detrital silicate materials (e.g., fel ds par, oli- vine, pyroxene, volcanic ash, and clays) to the marine environ ment where they can make up a significant portion of marine sedi ments. Here, recent work provides co mpelling geoche mical evidence for potentially substa ntial rates of silicate dissolutio n i n mari ne sedi me nts (e.g., Ki m et al., 2016; Ma her et al., 2004; Ma her et al., 2006; Sc holz et al., 2013; Solo mo n et al., 2008; Wall ma n n et al., 2008). T his is per haps u ns ur- prisi ng as sea water a nd pore waters are ge nerally u ndersaturated wit h respect to pri mary silicate mi nerals, and thus, their dissolution plays a role in buffering pore fluid p H and che mistry. Geoche mical evidence for mari ne weat heri ng deviates fro m t he traditio nal fra me work, w hic h specifi es t hat silicate mi neral disso- lution is largely li mited to terrestrial environ ments. Through the sa me Urey reaction path way invoked for co nti ne ntal silicate weat heri ng (eq uatio n (1)), t he ge neratio n of alkali nity a nd catio ns d uri ng t he weat heri ng of marine sedi ment plays a potentially large role in global carbon sequestration when coupled to carbonate for mation. Techniques used to fi ngerprint silicate mineral dissolution within sedi mentary syste ms include the applica- tio n of isoto pe tracers (e.g., 8 7 Sr/ 8 6 Sr, ( Ki m et al., 2016); 2 3 4 U/ 2 3 8 U ( Ma her et al., 2004; Ma her et al., 2006)) and the identifi cation of “ano malous” alkalinity or cation (e.g., Mg 2 + a n d C a 2 + ) fl uxes (Sc holz et al., 2013; Solo mon et al., 2008; Wall mann et al., 2008) within pore waters. Unexpectedly high alkalinity fluxes can

ISS O N ET AL. 1 4 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

be used as tracers for silicate weat heri ng by fi rst establis hi ng backgro u nd i n sit u pore water alkali nity levels t hro ug h t he applicatio n of diage netic reactio n tra nsport modeli ng (e.g., Sc holz et al., 2013; Solo mo n et al., 2008; Wall ma n n et al., 2008).

Silicate dissolution has been observed to take place throughout the sedi ment colu mn within virtually all redox zones (oxic, suboxic, sulfi dic, and methanic). Further, silicate weathering rates have been observed to evolve wit h dept h i nto t he sedi me nt pile a nd laterally bet wee n depositio nal e nviro n me nts. Bot h t he avail- ability of pri mary silicate mi nerals a nd pore water c he mistry ulti mately deter mi ne t he rate of silicate dissol u- tio n. Silicate mi neral availability varies depe ndi ng o n t he local tecto nic regi me, botto m water te mperature, a nd proxi mity to t he s horeli ne. For i nsta nce, i n sit u weat heri ng rates have bee n observed to vary by up to 4 orders in magnitude bet ween the deep sea site Ocean Drilling Progra m Site 984 in the North Atlantic and slope sedi ments on the Sakhalin Slope, Sea of Okhotsk ( Maher et al., 2006; Wall mann et al., 2008). Critically, pore water c he mistry evolves wit h dept h, most notably bet wee n redox zo nes tied to t he oxidatio n of orga nic matter, a nd has t he pote ntial to alter t he sat uratio n state (a nd t h us dissol utio n rate) of silicate phases. So me of the most rapid rates of silicate weathering have been observed to occur within the metha- noge nic zo ne of t he sedi me nt pile ( Wall ma n n et al., 2008). O n t he basis t hat met ha noge nesis is respo nsible

for the generation of ~5–20 T mol/year of C O 2 (and an equal a mount of C H4 ) globally ( Hi nric hs & Boetius, 2002; Reeburgh, 1993; Wall mann et al., 2012) and the assu mption that all this C O 2 is neutralized through silicate mi neral dissolutio n i n mari ne sedi me nts ( Wall ma n n et al., 2008), it has bee n proposed t hat roug hly a t hird of t his alkali nity is precipitated i n mari ne sedi me nts as a ut hige nic carbo nate, a nd t he rest fl uxed i nto overlying sea water ( Wall mann et al., 2008; Wall mann & Aloisi, 2012). 3.2.6. Weat heri ng of Ocea nic Crust Ocea nic cr ust t hat is co nti n ually accreted alo ng t he global mid ‐ocea n ridge net work covers more t ha n half of Eart h's s urface. T he upper most layer of t his cr ust is made up of hig h ‐porosity a nd per meability lavas t hro ug h which sea water continually circulates. The difference bet ween measured conductive heat flo w at the sea- floor and that predicted by ther mal models of the cooling of the oceanic lithosphere— the “missing heat” —indicates thatfluid flo w through the upper oceanic crust transports ~8–11 T W globally ( Hasterok, 2013; Stei n & Stei n, 1994). Fluid te mperatures wit hi n t he upper ocea nic crust are typically 5 –10 ° C hig her than that of botto m water during the first 10–20 Myr after crustal accretion when most fl uid‐rock reaction occurs ( Coogan & Gillis, 2018). At these te mperatures, fl uid fl uxes of ~0.6–1.7 × 10 1 6 kg/year (roughly 10– 50 % of the river flux) are required to transport the “missing heat ”. At modern sea water DI C levels of ~2.3 m mol/kg, this equates to ~1.5–4 × 10 1 3 moles of C passing through off ‐axis hydrother mal syste ms annually.

Fl uid ‐rock reactio ns wit hi n t he upper most cr ust lead to dissol utio n of basaltic p hases (e.g., glass, feldspar, a nd pyroxe ne) a nd precipitatio n of seco ndary mi nerals s uc h clays, zeolites, a nd calcite (or arago nite), provid- ing a sink for ocean DI C (carbon fro m the ocean‐at mosphere syste m). Because finding and sa mpling fluids exiti ng off‐axis hydrot her mal syste ms is tec h nically c halle ngi ng, most observatio nal co nstrai nts o n t he role of seafl oor weathering in the long‐ter m carbon cycle co me fro m studies of rocks altered in such syste ms. Staudigel et al. (1989) presented the fi rst detailed study of C uptake during seafl oor weathering focusing o n t hree closely spaced drill cores fro m t he wester n Atla ntic s ho wi ng t hat t he lavas i n t hese holes had take n 1 2 up substa ntial sea water C i n t he lava sectio n (~2.9 wt % C O 2 , eq uivale nt to ~2.9 × 10 mol/year if t hese were globally represe ntative). T he o nly way to have s uc h a large C uptake by t he cr ust (~10 % of t he C t hat passes through the crustal aquifer) appears to be through alkalinity generation during seafl oor weathering reac- tio ns ( Cooga n & Gillis, 2013; Spivack & Staudigel, 1994). Fra ncois a nd Walker (1992) proposed t hat t his sea- floor weathering C‐sink depended on botto m water p H and DI C and that it provided the do minant negative feedback on the long‐ter m carbon cycle. Ho wever, at the near‐neutral p H of sea water, the p H dependence of weat heri ng rates is li mited, leadi ng Caldeira (1995) to s uggest t hat t his feedback mec ha nis m was u nlikely to be effective.

No net heless, observatio ns fro m a larger nu mber of ocea n crust drill cores have no w s ho w n t hat t he total C O 2 uptake by t he lava sectio n of t he ocea nic cr ust t hro ug h ti me is variable a nd was sig ni fi ca ntly hig her for cr ust for med in the late Mesozoic than late Cenozoic ( Alt & Teagle, 1999; Gillis & Coogan, 2011). Further, most upper ocea nic crust carbo nate is precipitated wit hi n t he fi rst 20 Myr after crustal accretio n ( Cooga n et al.,

2016; Cooga n & Dosso, 2015; Staudigel & Hart, 1985), suggesti ng t hat t his differe nce i n C O 2 content refl ects

ISS O N ET AL. 1 5 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

so mething i ntri nsically different about t he syste m bet ween t hese ti mes— t he most likely candidates bei ng dif- ferences in ocean te mperature and co mposition. A significant te mperature dependence of seafloor basalt alteration was proposed by Brady and Gíslason (1997) based o n basalt‐sea water experi ments and is supported by modeling Sr‐isotope syste matics in these syste ms ( Coogan & Dosso, 2015) and by global C‐cycle models ( Krissansen‐Totton & Catling, 2017). Additionally, O ‐isotope ther mo metry sho ws that both the mini mu m and average te mperature of precipitation of carbo nate i n t he upper oceanic cr ust track botto m water te mpera- ture, being higher in the late Mesozoic than late Cenozoic ( Coogan & Gillis, 2018; Gillis & Coogan, 2011). T hese observatio ns lead to a si mple model for t he role of sea fl oor weat heri ng of basalts i n t he lo ng‐ter m carbo n

cycle in which increased at mospheric C O 2 gives rise to a n i ncreased global mea n s urface te mperat ures a nd i n tur n i ncrease i n botto m water te mperature of t he sa me mag nitude ( Krissanse n‐Totto n & Catli ng, 2017). T hus,

t he te mperature of sea water‐basalt reactio ns i n t he upper oceanic crust i ncreases leadi ng to elevated C O 2 c o n- su mption (alkali nity ge neration) a nd a stabilizing feedback o n t he C cycle. 3.3. I nter nal Carbo n Recycli ng ( Reverse Weatheri ng) To our k no wledge, t he notio n of reverse weathering (and marine weathering) can be traced back to a se minal paper by Sillé n (1961), w here clay mi nerals are proposed to act as a b uffer of sea water p H. I n t his co ntrib utio n a nd s ubseque nt p ublicatio ns, Sillé n q uestio ned t he traditio nal vie w of a carbo nate syste m operati ng as t he sole reg ulator of sea water p H, hig hlig hti ng t hat sig nifi cant a mo u nts of sea water alkali nity could be released ( mar- ine weathering) and/or consu med (reverse weathering) by silicate mineral phases (Sillén, 1961, 1967). M ultiple articles follo wed wit hi n t he decade, predo mi na ntly foc used o n hig hlig hti ng t he i mporta nce of reverse weathering for the budgets of multiple major cations, alkalinity and H + ( Garrels, 1965; Holland, 1965; Mackenzie & Garrels, 1965, 1966a, 1966b). Perhaps most notably, Mackenzie and Garrels constructed a mass balance for river water i nputs, suc h t hat t he major constitue nts were precipitated fro m sea water as mineral phases co m monly found in marine sedi ments— balancing their budgets required for Na, Mg, K and Si re moval through clay authigenesis ( Mackenzie & Garrels, 1966a, 1966b). Subsequently, both fi eld (e.g., Balder mann et al., 2013; Balder mann et al., 2015; Ehlert et al., 2016; Ku & Walter, 2003; Macke nzie et al., 1981; Macki n & Aller, 1984, 1986; März et al., 2015; Mic halopoulos et al., 2000; Michalopoulos & Aller, 1995, 2004; Presti & Michalopoulos, 2008; Rah man et al., 2016, 2017; Ristvet, 1978; Solo mon et al., 2008; Tatzel et al., 2015; Wall mann et al., 2008) and laboratory studies ( Loucaides et al., 2010; Mic halopoulos & Aller, 1995) i nvestigati ng t he mi neralogy a nd pore water c he mistry of modern marine sedi ments and their evolution with depth have provided convincing evidence for the operatio n of t his process i n t he natural e nviro n me nt, pro mpti ng greater accepta nce wit hi n t he broader co m- m u nity. Most rece ntly, st udies of stable Si ( E hlert et al., 2016), cos moge nic Si ( Ra h ma n et al., 2016, 2017) a nd berylliu m ( Be) ( Ber n hardt et al., 2020) isotopes have bee n developed as a tracer for fi ngerpri nti ng Si uptake during authigenic clay for mation in marine sedi ments. Si milar to marine weathering, reverse weathering has bee n observed to take place i n bot h mari ne sedi me nts (e.g., E hlert et al., 2016) a nd ocea nic cr ust (e.g., C ha n et al., 1992; C ha n et al., 2002; C ha n & Kast ner, 2000). Note t hat t he latter s ho uld not be co nf used wit h cation exchange bet ween fl uids and pri mary mineral phases in oceanic crust ( Berner & Berner, 2012), which have no direct effect o n t he global mari ne alkali nity bala nce. O n t he basis t hat silica for ms t he fra me work of all clay mi nerals, t he global silica cycle provides for a usef ul method of keeping track of reverse weathering fl uxes. Largely based on extrapolations fro m work carried out in the proxi mal A mazon delta ( Michalopoulos & Aller, 2004), dissolved silica sequestration as authigenic clay p hases globally was i nitially esti mated at ~1 to 1.5 T mol/year ( Holla nd, 2005; Laruelle et al., 2009; Tréguer & De La Rocha, 2013). For reference, the total a mount of ne w Si introduced into the ocean syste m is esti mated at ~10.9 T mol/year ( Tréguer & De La Rocha, 2013). Ho wever, recent work adopting the use of c os m o g e ni c 3 2 Si suggest that this fl ux could be significantly higher at ~4.5 to 4.9 T mol/year globally in coastal a nd deltaic syste ms ( Ra h ma n et al., 2017). I n t his novel co ntributio n, 3 2 Si is a p plie d as a tracer for st udyi ng Si release d uri ng bioge nic silica dissol utio n a nd its recapt ure by a ut hige nic p hases i n mari ne sedi- mentary pore waters ( Rah man et al., 2017). The capture of cos mogenic 3 2 Si (ge nerated by cos mic ray spalla- ti o n of 4 0 Ar i n t he at mosp here) by bioge nic silica (e.g., diato ms) upo n e nteri ng t he ocea n, delivers t he 3 2 Si t o t he sedi me nt‐water i nterface after t he orga nis m dies. Elevated 10 Be/9 Be ratios i n mari ne relative to riveri ne clay mi neral sedi me nt fractio ns provide f urt her s upporti ng evide nce for t he capt ure of Si (derived fro m bio- ge nic opal) d uri ng a ut hige nic clay for matio n. T hese esti mates do not acco u nt for (1) s helf, slope, up welli ng,

ISS O N ET AL. 1 6 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

and deep sea environ ments and (2) any a mount of Si capture that was derived fro m direct diffusion with marine botto m waters and/or Si released via dissolution of pri mary silicate minerals ( marine weathering) that do not host 3 2 Si a nd, t herefore, ca n be vie wed as co nservative. Reverse weat heri ng ca n i nvolve a variety of differe nt dissolved catio n species (e.g., Li + ,K+ , M g2 + , a n d F e2 + ), a nd so keepi ng track of t his process t hro ug h a ny of t hese i ndivid ual cycles will provide a n i nco mplete vie w of its i mpact o n t he global carbo n a nd alkali nity b udgets. Si milarly, t he global export of Si does not tra nslate directly to a fl ux of carbo n bei ng recycled t hroug h reverse weat heri ng. Ho wever, wit h k no wledge of t he alka- li nity ( Alk):silica (Si) co ns u mptio n ratio associated wit h clay a ut hige nesis, w hic h is deter mi ned by t he co m-

positio n (clay species) of t he globally i ntegrated clay mi neral asse mblage, t he reverse weat heri ng fl ux of C O 2 ca n be esti mated. A clay mi neral species/asse mblage wit h a hig her Alk:Si ratio will have a larger i mpact o n global carbo n recycli ng (Isso n & Pla navsky, 2018). T he for matio n of kaoli nite, for i nsta nce, does not co n- su me soluble catio ns a nd t hus alkali nity a nd so has no i mpact o n t he global carbo n cycle. Alk:Si ratios of standard clay co mpositions (note that many have fl exible stoichio metries) span a large range bet ween kao- li nite a nd up to 4.0. For i nsta nce, sepiolite has a n Al k:Si = 1.33, gree nalite = 3.0, bert hieri ne = 4.0. T he activ- ities of dissolve d catio ns, silica, H+ , a nd te mperat ure all play a role i n reg ulati ng t he sat uratio n state a nd t h us rate of for matio n of a n a ut hige nic clay mi neral p hase. T he depe nde nce ( A, B, a nd C) of eac h o ne is set by t he

stoic hio metry of t he clay mi neral (for i nsta nce, if X A Si B O 5 (OH)4 , and C = A for a monovalent cation and C = A × 2 for a divale nt catio n):

α A þ α B K ¼ c ati o n Si O 2 a qðÞ s p α C H þ

Organic matter re mineralization reactions have the potential to alter marine pore water che mistry and p H levels and their evolution with depth in the sedi ment pile, which have the potential to infl uence rates of reverse weathering (and for ward marine weathering). Hydrogen ion and alkalinity budgets evolve with depth in marine sedi ments. Specifically, the oxidation of organic matter tied to aerobic respiration and met ha noge nesis ge nerates acidity, w hile iro n a nd sulfate reductio n co nsu mes acidity. T hus, all else bei ng equal, t he c he mistry of t he upper portio n of a n a noxic mari ne sedi me nt pile above t he sulfate‐met ha ne tra n- sitio n zo ne would be expected to produce co nditio ns t hat favor reverse weat heri ng type reactio ns (i ncrease sat uratio n state of silicate mi nerals) a nd favor mari ne weat heri ng belo w it (sectio n 3.2.5).

4. Evol utio n of t he Global Carbo n Cycle 4.1. Co ntrols o n Degassi ng a nd its Evol utio n All of t he major carbo n dioxide o utgassi ng so urces ca n pote ntially vary dra matically t hro ug h ti me. T here are likely u nidirectio nal c ha nges i n t he history of o utgassi ng, tied to Eart h's lo ng‐ter m t her mal a nd tecto nic evo- lutio n a nd superi mposed o n t his s horter‐ter m cyclic c ha nges o n t he order of several millio n years li nked to tectonic processes. Historically, a long‐ter m drop in mantle heat fl o w is assu med to drive a progressive

decline in mid‐ocean ridge spreading rates and thus ridge C O 2 fl uxes. But this rapid spreading rate model clashes strongly with e mpirical observations and fi rst‐principle geophysical modeling, which provide strong evidence that the early Earth was instead characterized by slo w rates of ocean ridge spreading ( Korenaga, 2018). T his, ho wever, does not i mply t hat t here has bee n no decli ne i n o utgassi ng rates t hro ug h Eart h's his- tory. Most i mportantly, there is co mpelling evidence fro m modeling the 1 4 2 Nd isotope and zircon records that crustal recycling rates have declined through Earth's history— with the most precipitous decline in the Archean ( Korenaga, 2018; Rosas & Korenaga, 2018). These enhanced rates of continental crustal recy-

cling would have driven enhanced C O 2 outgassing in the Preca mbrian, providing part of the solution to the faint young Sun paradox. Rates of outgassing are also likely to have varied during “supercontinent” cycles which occur on a roughly 15 Myr ti me scale, and during mag matic fl are ups that typically occur on a several millio n‐year ti me scale or less (e.g., B urgess et al., 2014). T here is little debate t hat o utgassi ng fl uxes have likely varied t hro ug h Eart h history, b ut it is dif fi c ult to pro- vide co nstrai nts o n o utgassi ng evol utio n. T his is not s urprisi ng, give n t hat t here are large u ncertai nties eve n i n moder n o utgassi ngfl uxes (see sectio n 3.1). A n exceptio n to t his r ule is co nti ne ntal arc syste ms, t hat have

the potential to contribute large fl uxes of C O 2 to the at mosphere associated with the recycling of carbon

ISS O N ET AL. 1 7 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

preserved i n t he upper plate ( Lee et al., 2013; Lee & Lackey, 2015; Maso n et al., 2017). Detrital zirco n U ‐Pb ages extracted fro m clastic sedi mentary rocks provide a means to assess regional continental mag matis m in deep geologic ti me. Zircon is an accessory mineral co m monly produced in silicic mag mas (e.g., Lee & Bach mann, 2014), which are produced in large volu mes along continental arc syste ms. Ma fic rocks can cer- tai nly contai n zirco ns; ho wever, t hey are far less abu nda nt t ha n i n t he hig her silicic conte nt rocks. Zircons ca n re mai n i n upper crustal rocks for exte nded i ntervals of ti me beca use of t he mi neral's resilie nce to p hysical and che mical degradation (e.g., Gehrels, 2014). Thus, zircon can survive multiple episodes of sedi mentary recycli ng (i.e., b urial, exh u matio n, erosio n, a nd reb urial). T his explai ns w hy a sa ndstone may contai n a wide variety of zirco ns wit h crystallizatio n ages billio ns of years older t ha n t he depositional age of t he rock — eve n modern river sands can yield Archean aged zircon ( Ca mpbell & Allen, 2008). Clastic sedi ments deposited along conti ne ntal margi ns wit h arc syste ms, ho wever, te nd to contain large abu ndances of zirco n wit h crystal- lizatio n ages close to t he depositio nal age of t he sedi me nt ( Ca wood et al., 2012), w hic h are likely fi rst‐cycle zir- cons sourced fro m volcanic or rapidly exhu med pluto nic rocks produced by t he arc syste m. Sedi ment collected i n close proxi mity to t he arc will ge nerally be do mi nated by relatively yo u ng arc‐derived zirco ns, w hereas zir- con collected f urther a way fro m t he arc will likely contai n mixed zirco ns fro m older bedrock sources a nd wit h the abundance of the young grains being reduced ( Blu m & Pecha, 2014; Capaldi et al., 2017). The marine environ ment has changed dra matically with the advent of deep sea carbon burial in the Mesozoic

( Ridg well, 2005), and t his s hift i n t he carbonate factory could have c hanged C O2 o ut g assi n g. S p e ci fi c all y, t his would have i ncreased t he a mo unt of carbo nate o n average associated with s ubducting slabs globally, w hich (in

t he standard vie w) would i n tur n elevate t he a mount of carbo nate recycled a nd degassed as C O2 i n arc syste ms (e.g., Ed mond & H uh, 2003). I nteresti ngly, ho wever, t here is c urre ntly no evide nce for a step i ncrease i n s ur- face te mperatures associated wit h t he e merge nce of deep‐sea carbon burial, alt hough it s hould be noted t hat the Cretaceous and Cenozoic are ano malously war m and cool periods respectively. There is increasing evi-

dence for C O 2 degassing fro m metacarbonate rock in the overriding plate when there is fluid infiltration (Ste wart & Ague, 2018). In t his case, t he arc syste ms before a nd after t he o nset of extensive deep sea carbo n

burial may have, on average, had roughly si milar ratios of C O 2 outgassed to ne w zircons produced. Regardless, of t his u ncertai nty, it is reaso nable to ass u me t hat caref ul assess me nt of detrital zirco n age ‐data

may provide first‐order insights into global continental mag matis m and C O 2 outgassing through ti me. Rece nt st udies have utilized global detrital zirco n U ‐Pb data to track t he spatial distrib utio n of arc syste ms at vario us ti me scales ( Mc Ke nzie et al., 2014; Mc Ke nzie et al., 2016). A ge neral proble m i n st udies utilizi ng zir- con age co mpilations is the unevenness of data available, which biases co mposite global age distributions fro m being truly “global” —most distributions are biased to ward regions which contribute a substantial a mou nt of data ( Ca mpbell & Alle n, 2008; Voice et al., 2011). Mc Ke nzie et al. (2016) atte mpted to circu mve nt these sa mpling issues by binning their data by geographic region and depositional age, which allo wed the regional data to be nor malized prior to incorporation into te mporal co mposites. There are likely more sop histicated ways to avoid sa mpli ng biases, a nd as databases i ncrease t here will certai nly need to be f ut ure assess me nt of t hese records. But at prese nt, t his si mple nor malizatio n process provides a reaso nable mea ns of assessing ho w global zircon production has changed throughout ti me. T he sedi me ntary record of zirco n productio n fro m t he Cryoge nia n to t he prese nt (i.e., t he past ~720 millio n years of Eart h history) de mo nstrates a correlatio n wit h ice ho use‐gree n ho use tra nsitio ns. T his s uggests a ca u- sative link bet ween arc activity and cli mate transitions— when volcanis m was reduced Earth moved into an icehouse state, whereas when volcanic arcs expanded, Earth transitioned into a greenhouse cli mate ( Mc Kenzie et al., 2016). Correlation— even when there is an obvious mechanistic link— does not, necessitate causation. Ho wever, the current sedi mentary zircon record points to ward carbon outgassing history playing a critical role i n s hapi ng Eart h's cli mate. F ut ure work atte mpti ng to tease apart t he effects of prod uctio n ver- sus preservation on the sedi mentary zircon record and test the fra me work that this record can faithfully track arc activity would be of great i nterest to t he co m mu nity. Conclusions about outgassing fro m the zircon record can be tested to a degree by (1) considering ho w con- tinents have migrated and a malga mated throughout ti me (i.e., supercontinent breakup and for mation) and/or (2) mapping the kno wn distribution of ancient volcanic syste ms. Continental breakup requires the establish ment of rift syste ms and the opening of ne w ocean basins, which also requires the establish ment of subduction zones to acco m modate oceanic spreading. Therefore, as plates break up and migrate,

ISS O N ET AL. 1 8 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

volcanic arc syste ms will be more prevalent, whereas the collision of continental blocks shuts do wn and reduces arc syste ms. Extensive continental arc syste ms existed throughout the Ediacaran and early Paleozoic to for m Gond wana, during the mid‐Paleozoic to for m Pangea, and the Mesozoic to early Cenozoic follo wing the breakup of Pangea. These intervals correspond with greenhouse cli mates, whereas the final collisions that for med Gond wana, Pangea, and the closure of the Tethys ocean ( Ca wood & Buchan, 2007; Lee et al., 2013) all correspond with icehouses ( Mc Kenzie et al., 2014; Mc Kenzie et al., 2016). Mapping of arc syste ms in deep ti me is presently hindered by inconsistencies in reconstructing the extent and nature of mag matis m along ancient margin configurations and proble ms with pre‐Pangean plate co nfi guratio ns (e.g., Eva ns, 2013). No net heless, rece nt atte mpts to use paleogeograp hic maps to esti mate arc le ngt h a nd o utgassi ng have de mo nstrated si milar res ults to t hat of t he zirco n record ( Cao et al., 2017; Mills et al., 2017; Va n Der Meer et al., 2014). Esti mati ng t he areal exte nt of regio ns u ndergoi ng lo w‐, middle‐, a nd

high‐grade meta morphic alteration, which can also be a major C O 2 source (e.g., Ste wart & Ague, 2018; Ste wart et al., 2019; Z ha ng et al., 2018), is si milarly hi ndered by i nco nsiste nt geologic reco nstr uctio ns. B ut collectively, m ultiple li nes of data fro m rece nt st udies are de mo nstrati ng a co nsiste nt relatio ns hip bet wee n outgassi ng esti mates a nd major s hifts i n cli mate ( Cao et al., 2017; Mc Ke nzie et al., 2014; Mc Ke nzie et al., 2016; Mills et al., 2017; Va n Der Meer et al., 2014). T herefore, e mpirical records s upport t he idea t hat carbo n sources have played a major role i n co ntrolli ng Eart h's cli mate a nd provide a si mple expla natio n as to w hy Earth has toggled bet ween icehouse ‐greenhouse intervals. Large igneous provinces have been linked to short‐ter m cli mate shifts (fore most the end Per mian mass extinction and te mperature spike; Burgess & Bo wring, 2015) but cannot be linked to longer ‐ter m cli mate s wings and shifts given their li mited active lifes pa n. 4.2. Li nks Bet wee n t he Oxyge n, Iro n, Sulfur, a nd Carbo n Cycles T he iro n a nd s ulf ur cycles bot h have t he pote ntial to drive sig ni fi ca nt variatio ns i n at mosp heric carbo n diox- ide levels. T here are multiple burial c ha n nels for iro n a nd sulfur— sulfur ca n be buried as pyrite or sulfate evaporites, a nd iro n ca n be b uried as Fe clays, Fe carbo nates, pyrite, or oxides. At steady state, t he cycli ng of t hese ele me nts does not i mpact t he global carbo n cycle. O n t he ot her ha nd, s hifts i n t he iro n a nd sulfur

weat heri ng a nd burial ter ms ca n drive tra nsie nt yet geologically mea ni ngful s hifts i n pC O 2 tied to lo ng‐ter m i mbala nces i n t he global mari ne alkali nity b udget. T his idea b uilds o n t he sa me pri nciples i ntrod uced for sili- cate weathering and reverse weathering— nu merous iron and sulfur reactions either produce acidity or alka- linity. Sulfi de and ferrous iron oxidation produce acidity while ferric iron and sulfate reduction produce alkalinity— potentially altering the ocean‐at mosphere carbonic acid syste m balance and thus the ocean's capacity to hold i norga nic carbo n.

There are nu merous scenarios whereby changes in the iron and sulfur cycles could have driven a rise or decli ne i n at mosp heric carbo n dioxide levels. Ho wever, t here are t wo sce narios t hat have bee n t he most t hor- oughly explored and are most likely to have i mpacted Earth's cli mate on a several million ‐year ti me scale ( Bachan & Ku mp, 2015; Torres et al., 2014). An increase in pyrite weathering coupled to carbonate weath- ering can drive a ju mp in at mospheric carbon dioxide levels as long as the resulting sulfate accu mulates i n t he ocea n or is deposited as evaporite ( Torres et al., 2014). I n co ntrast, e n ha nced s ulfi de oxidatio n directly coupled to e n ha nced pyrite burial does not have a direct effect o n t he global carbo n cycle. T his tra nsie nt,

sulfur‐driven C O 2 source likely played an i mportant role in shaping Cenozoic cli mate, where there was a marked and progressive increase in marine sulfate concentrations (e.g., Wort mann & Paytan, 2012).

T he progressive oxidatio n of iro n i n t he upper co nti ne ntal crust ca n also result i n a net release of carbo n dioxide. Fore most, oxidatio n of co nti ne ntal cr ust ric h i n siderite a nd t he depositio n of sedi me nts ric h i n fer-

ric iron can result in a net C O2 release. Bachan and Ku mp (2015) proposed that this process was likely i mportant during the Archean‐Paleoproterozoic transition, which sa w Earth's first notable rise in at mo- spheric oxygen levels. Although there is strong evidence that the crust contained abundant Fe carbonates in the Archean— there is also abundant evidence for the deposition of Fe silicates (e.g., Johnson et al., 2018; Ras musse n et al., 2017) a nd Fe oxides (e.g., Bekker et al., 2014). Furt her, t he ocea ns appear to have re mained anoxic through most of the Preca mbrian and potentially even in the early Phanerozoic. Therefore, at mospheric oxygenation may not lead to a ju mp in ferric/ferrous ratio of marine sedi ments. Better constraints on the evolution of sedi mentary iron speciation are needed to further evaluate this proble m.

ISS O N ET AL. 1 9 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

Fig ure 10. Cartoon su m marizing the modern and Preca mbrian carbon cycle sources and sinks. Elevated rates of degassing have been predicted fro m geophysical models for t he Preca mbria n relative to t he moder n (e.g., Holla nd, 2009; O' Neill et al., 2014; Tajika & Mats ui, 1992), a nd o n t his basis carbo nate export tied to silicate weat heri ng must be proportio nally larger. Hig her dissolved silica a nd iro n levels t hat are hall mark features of t he Preca mbria n ocea n gave rise to co nditions t hat favored enhanced rates of reverse weathering, and da mpened marine weathering rates (Isson & Planavsky, 2018). By this fra me work, it follo ws that the ratio of terrestrial to mari ne silicate weat heri ng rates would have bee n hig her i n t he Preca mbria n.

4.3. Reverse Weathering Through Ti me Reverse weat heri ng could have evolved dra matically t hroug h ti me ( Figure 10). I n particular, it was rece ntly proposed t hat reverse weat heri ng rates were likely elevated i n early Preca mbria n ocea ns, directly li nked to t he hig her dissolved Si, Fe, a nd Mg levels t hat are traditio nally vie wed to be c haracteristic feat ures of t his ti me period (Isson & Planavsky, 2018). In this vie w, enhanced carbon recycling during this ti me could have

sustained a significantly elevated pC O 2 baseli ne, providi ng a solutio n to t he fai nt you ng Su n paradox t hat does not necessitate i nvoki ng a sig ni fi ca nt red uctio n to t he weat herability of Eart h's s urface. T he s ubseq ue nt radiation of siliceous organis ms (sponges, radiolarians, and most recently diato ms) fro m the latest Preca mbria n t hroug h t he P ha nerozoic forced a drop i n dissolved mari ne Si levels a nd t hus t he exte nt of car-

bon recycling through reverse weathering and in turn baseline pC O 2 levels. Marine authigenic clay mineral asse mblages are likely to have evolved through ti me. While authigenic clays have long been recognized to make up a signi ficant portion of the mineralogical and textural makeup of marine sedi ments (e.g., Hazen et al., 2013), a wide variety of species have bee n reported (for exa mple, palygorskite, mo nt morillo nite, gla u- co nite, sapo nite, bert hieri ne, gree nalite, a nd mi n nesotaite) ( Balder ma n n et al., 2015; e.g., B hattac haryya, 1983; Hazen et al., 2013; Johnson et al., 2018; Pletsch, 2001; Ras mussen et al., 2015; Ras mussen et al., 2013; Ras m usse n et al., 2017). For i nsta nce, Fe‐ric h clays wit h hig her Alk:Si co ns u mptio n ratios (e.g., gree n- alite, berthierine, and cha mosite) appear to be more abundant in anoxic Preca mbrian marine sedi ments than observed in the modern oceans characterized by lo wer Alk:Si consu mption ratios such as mont moril-

lonite. The drop in at mospheric pC O 2 levels since the Preca mbrian ( Kasting, 1987) can, to a signifi cant degree, be explai ned by t his evolutio n of clay mi neral asse mblage a nd Alk:Si.

5. F ut ure Directio ns

Although there have been major steps for ward in our understanding of the global carbon cycle in the past decade, many advances have resulted in more questions than ans wers. We make a case that marine pro- cesses ( marine weathering and reverse weathering) are likely more i mportant controls on long‐ter m cli mate t ha n traditio nally e nvisio ned (e.g., i n co ntrast to Ber ner, 2004). Ho wever, t here is a n obvio us need for better co nstrai nts o n t he exte nt a nd evol utio n of t hese processes. I n large part, t his will ce nter o n i mprovi ng o ur understanding of global cation mass balances. For the Mg cycle, this entails working to ward a better con- straint on the balance of clay to dolo mite for mation, and ho w the flux of Mg to the oceans has changed through ti me as the co mposition of the upper continental crust and weathering intensities have evolved. Mg isotopes provide a pro mising means to track these fluxes (e.g., Higgins & Schrag, 2015). Si milarly,

ISS O N ET AL. 2 0 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

rece nt work has hig hlig hted t he need for better esti mates o n t he exte nt of iro n burial t hat co nsu mes alkali- nity ( Balder ma n n et al., 2015; B hattac haryya, 1983; Ras m usse n et al., 2015; Tosca et al., 2016) a nd t here are poor e mpirical co nstrai nts o n ho w iro n fl uxes to t he ocea ns have c ha nged t hroug h ti me. T here is a need for a better basic u ndersta ndi ng of t he fate of Na a nd K i n t he ocea n, partic ularly co nstrai nts o n t he fractio n of these ele ments that are lost in alkalinity consu ming versus producing reactions (e.g., reverse weathering reactio ns verses seafl oor weat heri ng of feldspars). T he Ca cycle, wit h a si ngle o utp ut ter m, is typically co n- sidered to be t he best co nstrai ned catio n geoc he mical cycle. Ho wever, cou nter to t he sta ndard vie w, it has been suggested— and not yet refuted— that hydrother mal Ca fluxes do minate the Ca input to the oceans ( Caro et al., 2010). I n su m, t here are still major u ncertai nties i n t he sources a nd si nks of all major catio ns eve n i n t he moder n ocea ns, a nd t he evolutio n of catio n budgets t hroug h Eart h's history are eve n more poorly u nderstood. A mul- tifaceted approach— more well ‐grounded models, more experi mental data, and better constraints fro m the rock record— is needed to move for ward our understanding of evolution of cation budgets. Ho wever, an i ncrease i n t he a mo u nt of ki netic data for mari ne reactio ns (e.g., clay for matio n) mig ht lead to t he most sig- ni fi ca nt steps for ward i n t he s hort ter m. T he a mo u nt of ki netic data for co nditio ns releva nt to basalt altera- tio n a nd mari ne pore waters is well s hort of t hat for terrestrial soils. Better constraints on outgassing fl uxes are also essential to move for ward our understanding of the long ‐ ter m carbo n cycle. Alt houg h t here have bee n several rece nt large‐scale efforts (e.g., Trail by Fire project) to better co nstrai n volca nic o utgassi ng rates, several fl uxes are still hig hly spec ulative. Fore most, t he diff use outgassi ng fl ux is poorly co nstrai ned ( Kele me n & Ma n ni ng, 2015). T his s hould not be a surprise si nce diffuse fl uxes are of course more dif fi cult to constrain than point sources. Given the i mportance of fl uid infi ltration

in deter mining C O2 fluxes (Ste wart & Ague, 2018), more mechanistic models of fluid flo w in subduction zones and meta morphic terranes are also potentially pro mising targets. Lastly, exopla net researc h is rapidly developi ng as a fi eld a nd future space‐based telescope missio ns provide a strong i mpetus to translate our understanding of Earth's carbon cycle into a general theory for carbon cycli ng o n terrestrial pla nets. Movi ng to ward a more ge nerally accepted vie w of t he factors t hat have allo wed for Eart h to re mai n persiste ntly i n habited for billo n year ti me scales is a critical step to ward u ndersta ndi ng ho w terrestrial planets evolve. The “habitable zone ” for exoplanets— where liquid water can exist — is typi- cally delineated by assu ming the maxi mu m greenhouse gas capacity of a planet ( Kasting et al., 1993). T herefore, a well ‐grou nded ge neral t heory of carbo n cycli ng o n terrestrial pla nets is esse ntial to predict ho w likely a pla net wit hi n t he habitable zo ne will s ustai n t he evol utio n a nd persiste nce of life. F urt her, co u- pling astrono mical data with carbon cycle modeling may even help us pinpoint exoplanets most likely to harbor life (i.e., ideal target for exopla net at mosp heric c haracterizatio n).

6. S u m m a r y Eart h has bee n a habitable pla net for over 4 billio n years beca use of t he persiste nce of stabilizi ng feedbacks i n t he global carbo n a nd silico n cycles. Terrestrial silicate weat heri ng was traditio nally ass u med to give rise to t his negative feedback. I ndeed, work over t he last fe w decades s uggests t he prese nce of a terrestrial silicate weat heri ng feedback (e.g., Vo n Stra nd ma n n et al., 2013). Ho wever, t here are major u ncertai nties i n t he pro- cesses co ntrolli ng terrestrial silicate weat heri ng rates a nd t he bala nce bet wee n terrestrial a nd mari ne silicate weat heri ng. T he exte nt of silicate weat heri ng i n grou nd water has likely bee n u nderappreciated. Si milarly, marine weathering — both within the sedi ment pile and during oceanic crust alteration— is also likely a more i mportant part of the global carbon cycle than was traditionally envisioned ( Coogan & Gillis, 2018; e.g., Gillis & Cooga n, 2011; Wall ma n n et al., 2008). I n ot her words, terrestrial silicate weat heri ng i n t he cri- tical zo ne o nly acco u nts for part of t he carbo n re moval fro m t he ocea n‐at mosp here syste m a nd global silicate weathering fl uxes are much higher than those reconstructed fro m riverine alkalinity budgets. Ne w evidence for hig her moder n weat heri ng rates t ha n traditio nally e nvisio ned does not i mply t hat t he carbo n cycle is dra- matically out of steady state; carbo n source ter ms have likely also bee n u nderesti mated. Fore most, carbo n dioxide fl uxes fro m diffuse meta morphic outgassing are al most certainly higher than in traditional balanced lo ng‐ter m carbo n cycles budgets (e.g., Ber ner, 2004). Authigenic clay for mation in the oceans — reverse weathering— is a relatively minor part of the modern glo- bal carbo n cycle, b ut it is likely to have bee n more i mporta nt earlier i n Eart h's history (Isso n & Pla navsky,

ISS O N ET AL. 2 1 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

2018). T he s witc h to a biologically co ntrolled Si cycle— w here t he majority of Si O 2 fl uxes in the oceans are biogenic— caused a dra matic drop in dissolved marine Si concentrations (Siever, 1992). This drop in dissolved marine Si concentrations would have decreased the saturation state of clay minerals — leading to relatively less extentive reverse weathering and more intense marine weathering, and thus a decline in

baseline C O 2 levels due to less extensive carbon recycling within the ocean‐at mosphere syste m. Reverse weathering, the rates of which are p H dependent, also stabilizes the cli mate syste m — through the reverse weathering feedback. Si bio mineralization, a major biotic innovation, would have therefore destabilized Eart h's cli mate syste m, addi ng to o ne of t he ce ntral dog mas of t he origi nal Gaia hypot hesis, t hat life as a part of a global Earth syste m can give rise to both positive and negative cli mate feedbacks (Lovelock & Ku mp, 1994; Lovelock & Margulis, 1974). Give n t hat t here is still debate abo ut t he moder n carbo n cycle mass bala nce, it is u ns urprisi ng t hat t here is not a clear co nse nsus about t he factors t hat have drive n lo ng ‐ter m s hifts i n Eart h's cli mate. T his debate is often distilled do wn into argu ments about whether the sources or sinks of carbon sources have driven cli- mate s hifts. T here is a stro ng correlatio n bet wee n o utgassi ng tracers a nd cli mate records over t he last 800 millio n years (e.g., Mc Ke nzie et al., 2016; Mills et al., 2017), te ntatively s uggesti ng t hat carbo n so urces were the main drivers of cli mate oscillations. Ho wever, shifts in weatherability (carbon sinks) have also been linked to major cli mate shifts (e.g., Cenozoic cooling; Caves et al., 2016; Jagoutz et al., 2016; Zhang & Planavsky, 2019b). Resolving the factors controlling cli mate shifts and working to ward a more well ‐defi ned global carbo n mass bala nce are critical areas of f ut ure researc h.

Ackno wledg ments R ef e r e n c e s E. W. B. ackno wledges support fro m the Virtual Planetary Laboratory N AS A Ague, J. J. (2003). Fluid in fi ltration and transport of major, minor, and trace ele ments during regio nal meta morphis m of carbonate rocks, N AI. N. J. P. a nd T. T. I. ack no wledge Wepa waug Schist, Connecticut, US A. A merican Journal of Science ,303(9), 753–816. https://doi.org/10.2475/ajs.303.9.753 support fro m the Alternative Earths Ague, J. J., & Nicolescu, S. (2014). Carbon dioxide released fro m subduction zones by fl uid‐mediated reactions. Nature Geoscience ,7(5), N AS A N AI. L. R. K. is supported by the 355–360. https://doi.org/10.1038/ngeo2143 Heisi ng Si mo ns Fou ndatio n. J. J. A. a nd Algeo, T. J., Ber ner, R. A., May nard, J. B., & Sc heckler, S. E. (1995). Late Devo nia n ocea nic a noxic eve nts a nd biotic crises: “rooted” i n t he E. M. S. ackno wledge support fro m the evolution of vascular land plants. GS A Today ,5(3), 45–66. National Science Foundation ( E A R ‐ Aloisi, G., Wall mann, K., Dre ws, M., & Bohr mann, G. (2004). Evidence for the sub marine weathering of silicate minerals in Black Sea 1650329). The authors ackno wledge sedi ments: Possible i mplications for the marine Li and B cycles. Geoche mistry, Geophysics, Geosyste ms ,5, Q04007. https://doi.org/ discussio ns wit h G. E. Bebout, R. A. 10.1029/2003 G C000639 Ber ner, O. Beyssac, A. V. Brovaro ne, Alt, J. C., & Teagle, D. A. (1999). The uptake of carbon during alteration of ocean crust. Geochi mica et Cos mochi mica Acta ,63(10), C. P. Cha mberlain, T. Lyons, B. Marty, 1527–1535. https://doi.org/10.1016/S0016‐7037(99)00123‐4 F. Piccoli, D. Ru mble, C. Reinhard, M. Bachan, A., & Ku mp, L. R. (2015). The rise of oxygen and siderite oxidation during the Lo magundi Event. Proceedings of the National Sca mbell uri, C. M. Sc hiffries, a nd D. M. Acade my of Sciences ,112(21), 6562–6567. https://doi.org/10.1073/pnas.1422319112 Rye and also co m ments fro m C. ‐T. A. Balder ma n n, A., Warr, L., Letofsky ‐Papst, I., & Mavro matis, V. (2015). Substa ntial iro n sequestratio n duri ng gree n ‐clay aut hige nesis i n Lee and one other anony mous modern deep ‐sea sedi ments. Nature Geoscience ,8(11), 885–889. https://doi.org/10.1038/ngeo2542 r e vi e w er. Balder mann, A., Warr, L. N., Grathoff, G. H., & Dietzel, M. (2013). The rate and mechanis m of deep ‐sea glauconite for mation at the Ivory Coast –Ghana marginal ridge. Clays and Clay Minerals ,61(3), 258–276. https://doi.org/10.1346/ C C M N.2013.0610307 Becker, J. A., Bickle, M. J., Galy, A., & Holland, T. J. (2008). Hi malayan meta morphic C O2 fl uxes: Quantitative constraints fro m hydro- t her mal spri ngs. Earth and Planetary Science Letters ,265(3‐4), 616–629. https://doi.org/10.1016/j.epsl.2007.10.046 Bekker, A., Pla navsky, N., Ras m usse n, B., Krapez, B., Hof ma n n, A., Slack, J., et al. (2014). Iro n for matio ns: T heir origi ns a nd i mplicatio ns for ancient sea water che mistry. In Treatise on geoche mistry , ( Vol. 12, pp. 561–628). Net herla nds: Elsevier. Ber ner, E. K., & Ber ner, R. A. (2012). Global environ ment: Water, air, and geoche mical cycles . Pri nceto n: Pri nceto n U niversity Press. Berner, R. A. (1992). Weathering, plants, and the long ‐ter m carbon cycle.Geochi mica et Cos mochi mica Acta ,56(8), 3225–3231. https://doi. org/10.1016/0016‐7037(92)90300‐8 Berner, R. A. (1997). The rise of plants and their effect on weathering and at mospheric C O2. Science,276(5312), 544–546. https://doi.org/ 10.1126/scie nce.276.5312.544 Ber ner, R. A. (1998). T he carbo n cycle a nd carbo n dioxide over P ha nerozoic ti me: T he role of la nd pla nts. Philosophical Transactions of the Royal Society of London B: Biological Sciences ,353(1365), 75–82. https://doi.org/10.1098/rstb.1998.0192 Berner, R. A. (2004). The Phanerozoic carbon cycle: C O2 and O2 . Oxford: Oxford University Press. Berner, R. A., & Caldeira, K. (1997). The need for mass balance and feedback in the geoche mical carbon cycle. Geology ,25(10), 955–956. https://doi.org/10.1130/0091‐7613(1997)025 <0955:T NF MB A >2.3. C O;2 Ber ner, R. A., Lasaga, A. C., & Garrels, R. M. (1983). T he carbo nate ‐silicate geoc he mical cycle a nd its effect o n at mosp heric carbo n dioxide over t he past 100 millio n years. A merican Journal of Science ,283(7), 641–683. https://doi.org/10.2475/ajs.283.7.641 Ber n hardt, A., Oelze, M., Bouc hez, J., Bla ncke nburg, F., Mo htadi, M., C hristl, M., & Witt ma n n, H. (2020). 10 Be/ 9 Be ratios reveal mari ne authigenic clay for mation. Geophysical Research Letters . https://doi.org/10.1029/2019gl086061 B hattac haryya, D. P. (1983). Origi n of bert hieri ne i n iro nsto nes. Clays and Clay Minerals ,31(3), 173–182. https://doi.org/10.1346/ CC M N.1983.0310302 Bickle, M. (1996). Meta morp hic decarbonation, silicate weathering and the long ‐ter m carbo n cycle.Terra Nova ,8(3), 270–276. https://doi. org/10.1111/j.1365‐3121.1996.tb00756.x Blu m, M., & Pecha, M. (2014). Mid ‐Cretaceous to Paleocene North A merican drainage reorganization fro m detrital zircons. Geology ,42(7), 607–610. https://doi.org/10.1130/ G35513.1

ISS O N ET AL. 2 2 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

Blundy, J., Cash man, K. V., Rust, A., & Witha m, F. (2010). A case for C O2 ‐rich arc mag mas. Earth and Planetary Science Letters ,290(3‐4), 289–301. https://doi.org/10.1016/j.epsl.2009.12.013 Bluth, G. J., & Ku mp, L. R. (1994). Lithologic and cli matologic controls of river che mistry. Geochi mica et Cos mochi mica Acta ,58(10), 2341–2359. https://doi.org/10.1016/0016‐7037(94)90015‐9 Bluth, G. J. S., & Ku mp, L. R. (1991). Phanerozoic paleogeology. A merican Journal of Science ,291(3), 284–308. https://doi.org/10.2475/ ajs. 2 9 1. 3. 2 8 4 Bor mann, B. T., Wa ng, D., Snyder, M. C., Bor mann, F. H., Benoit, G., & April, R. (1998). Rapid, plant ‐induced weathering in an aggrading experi mental ecosyste m. Biogeoche mistry ,43(2), 129–155. https://doi.org/10.1023/ A:1006065620344 Brady, P. V., & Gíslason, S. R. (1997). Sea fl oor weathering controls on at mospheric C O2 and global cli mate. Geochi mica et Cos mochi mica Acta ,61(5), 965–973. https://doi.org/10.1016/S0016‐7037(96)00385‐7 Bra ntley, S. L. (2008). Ki netics of mi neral dissolutio n, Ki netics of water ‐rock i nteractio n, (pp. 151–210). Ne w York: Spri nger. B urgess, S. D., Bo wri ng, S., & S he n, S. Z. (2014). Hig h ‐precisio n ti meli ne for Eart h's most severe exti nctio n. Proceedi ngs of the Natio nal Acade my of Sciences ,111(9), 3316–3321. https://doi.org/10.1073/pnas.1317692111 Burgess, S. D., & Bo wring, S. A. (2015). High ‐precision geochronology confir ms volu minous mag matis m before, during, and after Earth's most severe extinction. Science Advances,1(7), e1500470. https://doi.org/10.1126/sciadv.1500470 Burnett, W. C., Bokunie wicz, H., Huettel, M., Moore, W. S., & Taniguchi, M. (2003). Ground water and pore water inputs to the coastal zone. Biogeoche mistry ,66(1/2), 3–33. https://doi.org/10.1023/ B: BI O G.0000006066.21240.53 Caldeira, K. (1995). Lo ng ‐ter m co ntrol of at mosp heric carbo n dioxide; lo w‐te mperature seafl oor alteratio n or terrestrial silicate‐rock weathering? A merican Journal of Science ,295(9), 1 0 7 7 – 1 1 1 4. Caldeira, K., & Kasti ng, J. F. (1992). T he life spa n of t he biosp here revisited. Nature ,360(6406), 721. Ca mpbell, I. H., & Allen, C. M. (2008). For mation of supercontinents linked to increases in at mospheric oxygen. Nature Geoscience ,1(8), 5 5 4. Cao, W., Lee, C. ‐T. A., & Lackey, J. S. (2017). Episodic nature of co nti ne ntal arc activity si nce 750 Ma: A global co mpilatio n. Earth and Planetary Science Letters ,461, 85–95. Capaldi, T. N., Horton, B. K., Mc Kenzie, N. R., Stockli, D. F., & Odlu m, M. L. (2017). Sedi ment provenance in contractional orogens The detrital zircon record fro m modern rivers in the Andean fold ‐thrust belt and foreland basin of western Argentina. Earth and Planetary Science Letters,479, 83–97. Caro, G., Papanastassiou, D., & Wasserburg, G. (2010). 40 K –40 Ca isotopic constraints on the oceanic calciu m cycle. Earth and Planetary Scie nce Letters,296(1‐2), 124–132. Caves, J. K., Jost, A. B., Lau, K. V., & Ma her, K. (2016). Ce nozoic carbo n cycle i mbala nces a nd a variable weat heri ng feedback. Earth and Planetary Science Letters ,450, 152–163. Ca wood, P. A., & Buchan, C. (2007). Linking accretionary orogenesis with supercontinent asse mbly. Earth ‐Science Revie ws,82(3‐4), 2 1 7 – 2 5 6. Ca wood, P. A., Ha wkes worth, C., & Dhui me, B. (2012). Detrital zircon record and tectonic setting. Geology ,40(10), 875–878. C ha mberli n, T. C. (1899). A n atte mpt to fra me a worki ng hypot hesis of t he cause of glacial periods o n a n at mosp heric basis. The Journal of Geology ,7(6), 545–584. Chan, L., Ed mond, J., Tho mpson, G., & Gillis, K. (1992). Lithiu m isotopic co mposition of sub marine basalts I mplications for the lithiu m cycle in the oceans. Earth and Planetary Science Letters ,108(1), 151–160. C ha n, L. ‐H., Alt, J. C., & Teagle, D. A. (2002). Lit hiu m a nd lit hiu m isotope pro fi les t hroug h t he upper ocea nic crust: a study of sea water– basalt exchange at O DP Sites 504 B and 896 A. Earth and Planetary Science Letters , 2 0 1 ( 1), 1 8 7 – 2 0 1. Chan, L. ‐H., & Kastner, M. (2000). Lithiu m isotopic co mpositions of pore fl uids and sedi ments in the Costa Rica subduction zone: I mplications for fl uid processes and sedi ment contribution to the arc volcanoes. Earth and Planetary Science Letters ,183(1), 275–290. Chavrit, D., Hu mler, E., & Grasset, O. (2014). Mapping modern C O2 fluxes and mantle carbon content all along the mid ‐ocean ridge syste m. Earth and Planetary Science Letters ,387, 229–239. Chen, Y., & Brantley, S. L. (1997). Te mperature ‐and p H‐dependence of albite dissolution rate at acid p H. Che mical Geology ,135(3‐4), 2 7 5 – 2 9 0. C hiodi ni, G., Fro ndi ni, F., Cardelli ni, C., Parello, F., & Per uzzi, L. (2000). Rate of diff use carbo n dioxide Eart h degassi ng esti mated fro m carbon balance of regional aquifers: The case of central Apenni ne, Italy. Journal of Geophysical Research,105( B4), 8423–8434. Cho, H. M., & Ki m, G. (2016). Deter mining ground water Ra end ‐me mber values for the esti mation of the magnitude of sub marine ground water discharge using Ra isotope tracers. Geophysical Research Letters ,43, 3865–3871. https://doi.org/10.1002/2016 G L068805 C ho, H. ‐M., Ki m, G., K wo n, E. Y., Moosdorf, N., Garcia ‐Orella na, J., & Sa ntos, I. R. (2018). Radiu m traci ng nutrie nt i nputs t hroug h sub- marine ground water discharge in the global ocean. Scientific Reports,8(1), 2439. Cooga n, L. A., & Dosso, S. E. (2015). Alteratio n of ocea n crust provides a stro ng te mperature depe nde nt feedback o n t he geological carbo n cycle a nd is a pri mary driver of t he Sr‐isotopic co mpositio n of sea water.Earth and Planetary Science Letters ,415, 38–46. https://doi.org/ 10.1016/j.epsl.2015.01.027 Cooga n, L. A., & Gillis, K. M. (2013). Evide nce t hat lo w ‐te mperature ocea nic hydrot her mal syste ms play a n i mporta nt role i n t he silicate‐ carbonate weathering cycle and long‐ter m cli mate regulation. Geoche mistry, Geophysics, Geosyste ms ,14, 1771–1786. https://doi.org/ 10.1002/ggge.20113 C o o g a n, L. A., & Gillis, K. M. ( 2 0 1 8). L o w ‐t e m perature alteration of the seafl oor: I mpacts on ocean che mistry. Annual Revie w of Earth and Planetary Sciences,46, 21–45. Cooga n, L. A., Parris h, R. R., & Roberts, N. M. (2016). Early hydrot her mal carbo n uptake by t he upper ocea nic cr ust: I nsig ht fro m i n sit u U ‐ Pb dating. Geology ,44(2), 147–150. D' A nto nio, M., Ibarra, D. E., & Boyce, C. K. (2019). La nd pla nt evolutio n decreased, rat her t ha n i ncreased, weat heri ng rates. Geology ,48(1), 2 9 – 3 3. Dasgupta, R. (2013). I ngassi ng, storage, a nd outgassi ng of terrestrial carbo n t hroug h geologic ti me. Revie ws in Mineralogy and Geoche mistry ,75(1), 183–229. Dasgupta, R., & Hirsch mann, M. M. (2006). Melting in the Earth's deep upper mantle caused by carbon dioxide. Nature ,440(7084), 659. Dasgupta, R., & Hirsch mann, M. M. (2010). The deep carbon cycle and melting in Earth's interior. Earth and Planetary Science Letters , 298(1‐2), 1–13. D' Errico, M. E., Lackey, J. S., S urpless, B. E., Loe wy, S. L., Woode n, J. L., Bar nes, J. D., et al. (2012). A detailed record of s hallo w hydro- ther malfluid flo w in the Sierra Nevada mag matic arc fro m lo w‐δ18 O skarn garnets. Geology ,40(8), 763–766. https://doi.org/10.1130/ G 3 3 0 0 8. 1

ISS O N ET AL. 2 3 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

Dole, S. (1964). Habitable planets for man , ( Vol. 1964). Ne w York: Blaisdell Pub: Co. Drever, J. I. (1994). T he effect of la nd pla nts o n weat heri ng rates of silicate mi nerals. Geochi mica et Cos mochi mica Acta ,58(10), 2325–2332. Duncan, M. S., & Dasgupta, R. (2014). C O2 solubility and speciation in rhyolitic sedi ment partial melts at 1.5 –3.0 GPa— I mplications for carbon flux in subduction zones. Geochi mica et Cos mochi mica Acta ,124, 328–347. Ed mond, J. M., & Huh, Y. (2003). Non ‐steady state carbonate recycling and i mplications for the evolution of at mospheric p C O2. Earth and Planetary Science Letters ,216(1‐2), 125–139. E hlert, C., Doeri ng, K., Wall ma n n, K., Sc holz, F., So m mer, S., Grasse, P., et al. (2016). Stable silico n isotope sig nat ures of mari ne pore waters — Biogenic opal dissolution versus authigenic clay mineral for mation. Geochi mica et Cos mochi mica Acta ,191, 102–117. Evans, D. A. (2013). Reconstructing pre ‐Pangean supercontinents. Bulletin ,125(11‐12), 1735–1751. Ferry, J. M. (1988). In filtration‐driven meta morphis m in northern Ne w England, US A. Journal of Petrology,29(6), 1121–1159. France ‐Lanord, C., & Derry, L. A. (1997). Orga nic carbo n burial forcing of the carbo n cycle fro m Hi malayan erosion. Nature ,390(6655), 65. Fra ncois, L. M., & Walker, J. C. (1992). Modelli ng t he P ha nerozoic carbo n cycle a nd cli mate: co nstrai nts fro m t he 87Sr/86Sr isotopic ratio of sea water. A merican Journal of Science ,292(2), 81–135. Gaillardet, J., Dupré, B., Louvat, P., & Allegre, C. (1999). Global silicate weat heri ng a nd C O2 co nsu mptio n rates deduced fro m t he che mistry of large rivers. Che mical Geology ,159(1), 3–30. Galy, V., Fra nce ‐La nord, C., Beyssac, O., Faure, P., Kudrass, H., & Pal hol, F. (2007). Ef fi cie nt orga nic carbo n burial i n t he Be ngal fa n sustained by the Hi malayan erosional syste m. Nature ,450(7168), 407. Galy, V., France ‐Lanord, C., Beyssac, O., Lartiges, B., & R ha ma n, M. (2010). Organic carbon cycling during Hi malayan erosion: Processes, fluxes and consequences for the global carbo n cycle , (pp. 163–181). Cli mate C ha nge a nd Food Security i n Sout h Asia, Dordrec ht: Spri nger. Garrels, R. M. (1965). Silica: Role i n t he b ufferi ng of nat ural waters. Scie nce,148(3666), 69–69. Gehrels, G. (2014). Detrital zircon U ‐Pb geochronology applied to tecto nics. Annual Revie w of Earth and Planetary Sciences ,42, 127–149. Gibling, M. R., & Davies, N. S. (2012). Palaeozoic landscapes shaped by plant evolution. Nature Geoscience ,5(2), 99. Gillis, K., & Cooga n, L. (2011). Sec ular variatio n i n carbo n uptake i nto t he ocea n cr ust. Earth a nd Pla netary Scie nce Letters ,302(3), 385–392. Gor man, P. J., Kerrick, D., & Connolly, J. (2006). Modeling open syste m meta morphic decarbonation of subducting slabs. Geoche mistry, Geophysics, Geosyste ms ,7, Q04007. https://doi.org/10.1029/2005 G C001125 Go ug h, D. (1981). Solar i nterior structure a nd lu mi nosity variatio ns , (pp. 21–34). P hysics of Solar Variatio ns, Dordrec ht: Spri nger. Grassi, D., & Sch midt, M. W. (2011). The melting of carbonated pelites fro m 70 to 700 k m depth. Journal of Petrology,52(4), 765–789. Grif fi t hs, R., Ba ha m, J., & Cald well, B. (1994). Soil sol utio n c he mistry of ecto mycorr hizal mats i n forest soil. Soil Biology a nd Bioche mistry, 26(3), 331–337. Hart, M. H. (1978). The evolution of the at mosphere of the Earth. Icarus,33(1), 23–39. Hart mann, J., & Moosdorf, N. (2012). The ne w global lithological map database G Li M: A representation of rock properties at the Earth surface. Geoche mistry, Geophysics, Geosyste ms ,13, Q12004. https://doi.org/10.1029/2012 G C004370 Hart mann, J., Moosdorf, N., Lauer wald, R., Hi nderer, M., & West, A. J. (2014). Global che mical weathering and associated P ‐release— The role of lithology, te mperature and soil properties. Che mical Geology ,363, 145–163. https://doi.org/10.1016/j.che mgeo.2013.10.025 Hasterok, D. (2013). Global patterns and vigor of ventilated hydrother mal circulation through young sea fl oor. Earth and Planetary Science Letters,380, 12–20. Haze n, R. M., Sverje nsky, D. A., Azzoli ni, D., Bis h, D. L., El more, S. C., Hi n nov, L., & Millike n, R. E. (2013). Clay mi neral evol utio n. A merican Mineralogist ,98(11‐12), 2 0 0 7 – 2 0 2 9. Helo, C., Lo ngpré, M. ‐A., S hi mizu, N., Clague, D. A., & Stix, J. (2011). Explosive eruptio ns at mid ‐ocea n ridges drive n by C O2‐ric h mag mas. Nature Geoscience ,4(4), 260. Higgins, J. A., & Schrag, D. P. (2015). The Mg isotopic co mposition of Ce nozoic sea water — Evidence for a link bet ween Mg ‐clays, sea water Mg/ Ca, and cli mate. Earth and Planetary Science Letters ,416, 73–81. Hilto n, D. R., Fisc her, T. P., & Marty, B. (2002). Noble gases a nd volatile recycli ng at subd uctio n zo nes. Revie ws in Mineralogy and Geoche mistry ,47(1), 319–370. Hinrichs, K. ‐U., & Boetius, A. (2002). The anaerobic oxidation of methane: Ne w insights in microbial ecology and biogeoche mistry. In Ocean margin syste ms , (pp. 457–477). Berlin, Heidelberg: Springer. Hirsch mann, M. M. (2018). Co mparative deep Earth volatile cycles: The case for C recycling fro m exosphere/ ma ntle fractionation of major ( H2 O, C, N) volatiles and fro m H2 O/ Ce, C O2/ Ba, and C O2/ Nb exosphere ratios. Earth and Planetary Science Letters ,502, 262–273. Hoff ma n, P. F., Abbot, D. S., As hke nazy, Y., Be n n, D. I., Brocks, J. J., Co he n, P. A., et al. (2017). S no wball Eart h cli mate dy na mics a nd Cryogenian geology ‐geobiology. Science Advances,3(11), e1600983. https://doi.org/10.1126/sciadv.1600983 Högbo m, A. (1894). On the probability of secular variations of at mospheric carbon dioxide (in S wedish). Svensk Ke misk Tidskrift,6, 1 6 9 – 1 7 6. Holla nd, H. D. (1965). T he history of ocea n water a nd its effect o n t he c he mistry of t he at mosp here. Proceedings of the National Acade my of Sciences,53(6), 1173–1183. Holland, H. D. (1984). The che mical evolution of the at mosphere and oceans . Princeton: Princeton University Press. Holland, H. D. (2002). Volca nic gases, black s mokers, and the Great Oxidation Event. Geochi mica et Cos mochi mica Acta ,66(21), 3811–3826. Holland, H. D. (2005). Sea level, sedi ments and the co mposition of sea water. A merican Journal of Science ,305(3), 220–239. Holland, H. D. (2009). Why the at mosphere beca me oxygenated: A proposal. Geochi mica et Cos mochi mica Acta ,73(18), 5241–5255. Hoyle, F. (1957). The black cloud, (pp. 26–27). London: Penguin. Ilyi nskaya, E., Mobbs, S., B urto n, R., B urto n, M., Pardi ni, F., Pfeffer, M. A., et al. (2018). Globally sig nifi ca nt C O2 e missio ns fro m Katla, a subglacial volcano in Iceland. Geophysical Research Letters ,45, 10,332–10,341. https://doi.org/10.1029/2018 G L079096 Isso n, T. T., & Pla navsky, N. J. (2018). Reverse weat heri ng as a lo ng‐ter m stabilizer of mari ne p H a nd pla netary cli mate. Nature ,560(7719), 4 7 1 – 4 7 5. Jagoutz, O., Macdo nald, F. A., & Royde n, L. (2016). Lo w‐latitude arc–co nti ne nt collisio n as a driver for global cooli ng. Proceedings of the National Acade my of Sciences ,113(18), 4935–4940. Jo h nso n, J. E., Mu hli ng, J. R., Cos midis, J., Ras m usse n, B., & Te mpleto n, A. S. (2018). Lo w‐Fe (III) gree nalite was a pri mary mi neral fro m Neoarchean Oceans. Geophysical Research Letters ,45, 3182–3192. https://doi.org/10.1002/2017 GL076311 Kasting, J. F. (1987). Theoretical constraints on oxygen and carbon dioxide concentrations in the Preca mbrian at mosphere. Preca mbrian research,34(3), 205–229. Kasti ng, J. F. (2019). T he Goldilocks pla net? Ho w silicate weat heri ng mai ntai ns Eart h “just rig ht”.Ele ments: An International Magazine of Mineralogy, Geoche mistry, and Petrology ,15(4), 235–240. Kasting, J. F., Whit mire, D. P., & Reynolds, R. T. (1993). Habitable zones arou nd main sequence stars. Icarus,101(1), 108–128.

ISS O N ET AL. 2 4 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

Kele men, P. B., Hanghøj, K., & Greene, A. (2003). One vie w of the geoche mistry of subduction ‐related mag matic arcs, with an e mphasis on pri mitive andesite and lo wer crust. Treatise on geoche mistry ,3, 659. Kele men, P. B., & Manning, C. E. (2015). Reevaluating carbon fl uxes in subduction zones, what goes do wn, mostly co mes up. Proceedings of the National Acade my of Sciences,112(30), E3997–E4006. https://doi.org/10.1073/pnas.1507889112 Keller, C., & Wood, B. (1993). Possibility of che mical weathering before the advent of vascular land plants. Nature ,364(6434), 223–225. Kerrick, D., & Caldeira, K. (1993). Paleoat mospheric consequences of C O2 released during early Cenozoic regional meta morphis m in the Tethyan orogen. Che mical Geology ,108(1‐4), 201–230. Kerrick, D., & Caldeira, K. (1999). Was the Hi malayan orogen a cli matically signi fi cant coupled source and sink for at mospheric C O2 during the Cenozoic? Earth and Planetary Science Letters ,173(3), 195–203. Kerrick, D., & Co n nolly, J. (2001). Meta morp hic devolatilizatio n of subd ucted mari ne sedi me nts a nd t he tra nsport of volatiles i nto t he Earth's mantle. Nature ,411(6835, 293. Kerrick, D. M., & Caldeira, K. (1998). Meta morphic C O2 degassing fro m orogenic belts. Che mical Geology ,145(3‐4), 213–232. Ki m, J. H., Torres, M. E., Haley, B. A., Ry u, J. S., Park, M. H., Ho ng, W. L., & C hoi, J. (2016). Mari ne silicate weat heri ng i n t he a noxic sedi ment of the Ulleung Basin: Evidence and consequences. Geoche mistry, Geophysics, Geosyste ms ,17, 3437–3453. https://doi.org/ 10.1002/2016 G C006356 Korenaga, J. (2018). Crustal evolution and mantle dyna mics through Earth history. Philosophical Transactions of the Royal Society A , 376(2132). https://doi.org/10.1098/rsta.2017.0408 Krissa nse n ‐Totto n, J., & Catli ng, D. C. (2017). Co nstrai ni ng cli mate se nsitivity a nd co nti ne ntal vers us sea fi oor weat heri ng usi ng a n i nverse geological carbon cycle model. Nature ,8(15423), 1. Ku, T., & Walter, L. (2003). Sy ndepositio nal for matio n of Fe ‐ric h clays i n tropical s helf sedi me nts, Sa n Blas Arc hipelago, Pa na ma. Che mical Geology ,197(1‐4), 197–213. Ku mp, L. R. (2018). Prolo nged Late Per mia n ‐Early Triassic hypert her mal: Failure of cli mate regulatio n? Philosophical Transactions. Series A, Mathe matical, Physical, and Engineering Sciences , 3 7 6 ( 2 1 3 0). https://doi.org/10.1098/rsta.2017.0078 Ku mp, L. R., & Arthur, M. A. (1997). Global che mical erosion during the Cenozoic: Weatherability balances the budgets. In Tectonic Uplift and Cli mate Change, (pp. 399–426). Boston, M A: Springer. Ku mp, L. R., Brantley, S. L., & Arthur, M. A. (2000). Che mical weathering, at mospheric C O2, and cli mate. Annual Revie w of Earth and Planetary Sciences ,28(1), 611–667. K wo n, E. Y., Ki m, G., Pri mea u, F., Moore, W. S., C ho, H. M., De Vries, T., et al. (2014). Global esti mate of s ub mari ne gro u nd water disc harge based on an observationally constrained radiu m isotope model. Geophysical Research Letters ,41, 8438–8444. https://doi.org/10.1002/ 2014 GL061574 Laakso, T. A., & Schrag, D. P. (2014). Regulation of at mospheric oxygen during the Proterozoic. Earth and Planetary Science Letters ,388, 81–91. https://doi.org/10.1016/j.epsl.2013.11.049 Lar uelle, G. G., Ro ubeix, V., Sferratore, A., Brod herr, B., Ci uffa, D., Co nley, D., et al. (2009). A nt hropoge nic pert urbatio ns of t he silico n cycle at t he global scale: Key role of t he la nd‐ocea n tra nsitio n. Global Biogeoche mical Cycles ,23, G B4031. https://doi.org/10.1029/ 2008 GB003267 Le Voyer, M., Kelley, K. A., Cottrell, E., & Hauri, E. (2017). Heteroge neity i n ma ntle carbo n co nte nt fro m C O 2 ‐u ndersaturated basalts. Nature Co m munications ,8(1), 14,062. https://doi.org/10.1038/nco m ms14062 Lee, C. ‐T. A., & Bac h ma n n, O. (2014). Ho w i mporta nt is t he role of crystal fractio natio n i n maki ng i nter mediate mag mas? I nsig hts fro m Zr and P syste matics. Earth and Planetary Science Letters ,393, 266–274. Lee, C. ‐T. A., & Lackey, J. S. (2015). Global co nti ne ntal arc fl are‐ups a nd t heir relatio n to lo ng ‐ter m gree n house co nditio ns.Ele ments ,11(2), 1 2 5 – 1 3 0. Lee, C. ‐T. A., S he n, B., Slot nick, B. S., Liao, K., Dicke ns, G. R., Yokoya ma, Y., et al. (2013). Co nti ne ntal arc –isla nd arcfl uct uatio ns, gro wt h of crustal carbonates, and long‐ter m cli mate change. Geosphere ,9(1), 21–36. Lee, C. ‐T. A., T h ur ner, S., Paterso n, S., & Cao, W. (2015). T he rise a nd fall of co nti ne ntal arcs: I nterplays bet wee n mag matis m, uplift, weathering, and cli mate. Earth and Planetary Science Letters ,425, 105– 1 1 9. L e nt o n, T. M. (2001). The role of land plants, phosphorus weathering and fi re in the rise and regulation of at mospheric oxygen. Global Change Biology ,7(6), 613–629. Li, G., & Elder fi eld, H. (2013). Evolution of carbo n cycle over the past 100 million years. Geochi mica et Cos mochi mica Acta ,103, 11–25. Li, K. ‐F., Pa hleva n, K., Kirsc hvi nk, J. L., & Y u ng, Y. L. (2009). At mosp heric press ure as a nat ural cli mate reg ulator for a terrestrial pla net with a biosphere. Proceedings of the National Acade my of Sciences ,106(24), 9576–9579. Li, L., Ma her, K., Navarre ‐Sitc hler, A., Dr u ha n, J., Meile, C., La wre nce, C., et al. (2017). Expa ndi ng t he role of reactive tra nsport models i n critical zone processes. Earth ‐Science Revie ws,165, 280–301. Li, Y. ‐H. (2000). A co mpendiu m of geoche mistry: Fro m solar nebula to the hu man brain . Princeton: Princeton University Press. Liu, Y., Ya manaka, T., Zhou, X., Tian, F., & Ma, W. (2014). Co mbined use of tracer approach and nu merical si mulation to esti mate ground water recharge in an alluvial aquifer syste m: A case study of Nasunogahara area, central Japan. Journal of Hydrology,519, 8 3 3 – 8 4 7. Lo ucaides, S., Mic halopo ulos, P., Presti, M., Ko ni ng, E., Be hre nds, T., & Va n Cappelle n, P. (2010). Sea water ‐mediated i nteractio ns bet wee n diato maceous silica and terrigenous sedi ments: Results fro m long‐ter m incubation experi ments. Che mical Geology ,270(1‐4), 68–79. Lovelock, J. E., & Ku mp, L. R. (1994). Failure of cli mate regulation in a geophysiological model. Nature ,369(6483), 732. Lovelock, J. E., & Margulis, L. (1974). At mospheric ho meostasis by and for the biosphere: The Gaia hypothesis. Tellus ,26(1‐2), 2–10. Lovelock, J. E., & Whit fi eld, M. (1982). Life span of the biosphere. Nature ,296(5857), 561–563. Lo we nstei n, T. K., Hardie, L. A., Ti mofeeff, M. N., & De micco, R. V. (2003). Secular variatio n i n sea water c he mistry a nd t he origi n of calciu m chloride basinal brines. Geology ,31(10), 857–860. Macke nzie, F., Ristvet, B., T horste nso n, D., Ler ma n, A., a nd Leeper, R., 1981, Reverse weat heri ng a nd c he mical mass bala nce i n a coastal e n vir o n m e n t. Mackenzie, F. T., & Garrels, R. (1971). Evolution of sedi mentary rocks . Ne w York: Norton. Macke nzie, F. T., & Garrels, R. M. (1965). Silicates: reactivity wit h sea water. Scie nce,150(3692), 57–58. Mackenzie, F. T., & Garrels, R. M. (1966a). Che mical mass balance bet ween rivers and oceans. A merican Journal of Science , 264(7), 5 0 7 – 5 2 5. Macke nzie, F. T., & Garrels, R. M. (1966b). Silica ‐bicarbo nate bala nce i n t he ocea n a nd early diage nesis. Journal of Sedi mentary Petrology, 3 6 ( 4).

ISS O N ET AL. 2 5 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

Macki n, J. E., & Aller, R. C. (1984). Dissolved Al i n sedi me nts a nd waters of t he East C hi na Sea: I mplicatio ns for a ut hige nic mi neral for- mation. Geochi mica et Cos mochi mica Acta ,48(2), 281–297. Macki n, J. E., & Aller, R. C. (1986). T he effects of clay mi neral reactio ns o n dissolved Al distrib utio ns i n sedi me nts a nd waters of t he A mazon continental shelf. Continental Shelf Research ,6(1), 245–262. Maher, K., & Cha mberlain, C. (2014). Hydrologic regulation of che mical weathering and the geologic carbon cycle. Science,343(6178), 1 5 0 2 – 1 5 0 4. Ma her, K., De Paolo, D. J., & Li n, J. C. ‐F. (2004). Rates of silicate dissol utio n i n deep ‐sea sedi me nt: I n sit u meas ure me nt usi ng 234 U/238 U of pore fluids. Geochi mica et Cos mochi mica Acta ,68(22), 4629–4648. Ma her, K., Steefel, C. I., De Paolo, D. J., & Via ni, B. E. (2006). T he mi neral dissol utio n rate co n u ndr u m: I nsig hts fro m reactive tra nsport modeling of U isotopes and pore fluid che mistry in marine sedi ments. Geochi mica et Cos mochi mica Acta ,70(2), 337–363. Marty, B., & Tolstikhin, I. N. (1998). C O2 fl uxes fro m mid‐ocean ridges, arcs and plu mes. Che mical Geology ,145(3), 233–248. März, C., Meinhardt, A. ‐K., Schnetger, B., & Bru msack, H. ‐J. (2015). Silica diagenesis and benthic fl uxes in the Arctic Ocean. Marine Che mistry ,171, 1–9. Maso n, B., a nd Moore, C., 1982, Co nce ntratio ns i n crust, gra nite, Diabase (basalt), a nd s hale: Pri nciples of Geoc he mistry, p. 46 ‐47. Mason, E., Ed monds, M., & Turchyn, A. V. (2017). Re mobilization of crustal carbon may do minate volcanic arc e missions. Science, 357(6348), 290–294. Matt he ws, S., S horttle, O., Rudge, J. F., & Macle n na n, J. (2017). Co nstrai ni ng ma ntle carbo n: C O2 ‐trace ele me nt syste matics i n basalts a nd the roles of mag ma mixing and degassing. Earth and Planetary Science Letters ,480, 1–14. https://doi.org/10.1016/j.epsl.2017.09.047 Mc Ke nzie, N. R., Horto n, B. K., Loo mis, S. E., Stockli, D. F., Pla navsky, N. J., & Lee, C. ‐T. A. (2016). Co nti ne ntal arc volca nis m as t he pri ncipal driver of ice house‐gree n house variability. Science, 3 5 2 ( 6 2 8 4), 4 4 4 – 4 4 7. Mc Ke nzie, N. R., Hug hes, N. C., Gill, B. C., & Myro w, P. M. (2014). Plate tecto nic i n fl ue nces o n Neoproterozoic –early Paleozoic cli mate a nd ani mal evolution. Geology ,42(2), 127–130. Meybeck, M. (1987). Global che mical weathering of sur fi cial rocks esti mated fro m river dissolved loads. A merican Journal of Science , 287(5), 401–428. https://doi.org/10.2475/ajs.287.5.401 Michalopoulos, P., & Aller, R. C. (1995). Rapid clay mineral for mation of A mazon delta sedi ments: Reverse weathering and oceanic ele- mental cycles. Science,270(5236), 614. Mic halopo ulos, P., & Aller, R. C. (2004). Early diage nesis of bioge nic silica i n t he A mazo n delta: Alteratio n, a ut hige nic clay for matio n, a nd storage. Geochi mica et Cos mochi mica Acta ,68(5), 1061–1085. Mic halopo ulos, P., Aller, R. C., & Reeder, R. J. (2000). Co nversio n of diato ms to clays d uri ng early diage nesis i n tropical, co nti ne ntal s helf muds. Geology ,28(12), 1095–1098. Milli ma n, J. D., & Far ns wort h, K. L. (2011). R u noff, erosio n, a nd delivery to t he coastal ocea n. River discharge to the coastal ocea n: a global synthesis (pp.13–69). Ca mbridge, U K: Ca mbridge University Press. Mills, B. J., Scotese, C. R., Waldi ng, N. G., S hields, G. A., & Le nto n, T. M. (2017). Elevated C O2 degassi ng rates preve nted t he ret ur n of Sno wball Earth during the Phanerozoic. Nature Co m munications ,8(1), 1110. Misra, S., & Froelic h, P. N. (2012). Lit hiu m isotope history of Ce nozoic sea water: C ha nges i n silicate weat heri ng a nd reverse weat heri ng. Science,335(6070), 818–823. Mojzsis, S. J., Harriso n, T. M., & Pidgeo n, R. T. (2001). Oxyge n ‐isotope evide nce fro m a ncie nt zirco ns for liq uid water at t he Eart h's s urface 4,300 Myr ago. Nature ,409(6817), 178. Moore, W. S., Sar mie nto, J. L., & Key, R. M. (2008). Sub mari ne grou nd water disc harge revealed by 228 Ra distributio n i n t he upper Atla ntic Ocean. Nature Geoscience ,1(5), 309–311. https://doi.org/10.1038/ngeo183 Moosdorf, N., & Oehler, T. (2017). Societal use of fresh sub marine ground water discharge: An overlooked water resource. Earth ‐Science Revie ws ,171, 338–348. Moulto n, K. L., & Ber ner, R. A. (1998). Qua nti fi catio n of t he effect of pla nts o n weat heri ng: Studies i n Icela nd. Geology ,26(10), 895–898. Mulligan, A. E., & Charette, M. A. (2009). Ground water flo w to t he coastal ocea n: Ele me nts of p hysical ocea nograp hy. A derivative of the encyclopedia of ocean sciences,465. Nesbitt, B. E., Me ndoza, C. A., & Kerrick, D. M. (1995). Surface fl uid convection during Cordilleran extension and the generation of meta morphic C O2 contributions to Cenozoic at mospheres. Geology ,23(2), 99–101. O' Neill, C., Le nardic, A., Höi nk, T., a nd Coltice, N., 2014, Ma ntle co nvectio n a nd outgassi ng o n terrestrial pla nets, Co mparative Cli matology of Terrestrial Pla nets, 473 p.: Ozaki, K., Rei n hard, C. T., & Tajika, E. (2019). A sluggis h mid ‐Proterozoic biosp here a nd its effect o n Eart h's redox bala nce. Geobiology , 17(1), 3–11. Paytan, A., Shellenbarger, G. G., Street, J. H., Go nneea, M. E., Davis, K., Young, M. B., & Moore, W. S. (2006). Sub marine grou nd water discharge: an i mportant source of ne w inorganic nitrogen to coral reef ecosyste ms. Li mnology and Oceanography ,51(1), 343–348. C hi c a g o Pen man, D. E. (2016). Silicate weathering and North Atlantic silica burial during the Paleocene ‐Eocene Ther mal Maxi mu m. Geology ,44(9), 7 3 1 – 7 3 4. Piccoli, F., Brovaro ne, A. V., Beyssac, O., Marti nez, I., Ag ue, J. J., & C had utea u, C. (2016). Carbo natio n by fl uid –rock i nteractio ns at hig h‐ pressure conditions: I mplications for carbon cycling in subduction zones. Earth and Planetary Science Letters ,445, 146–159. Plank, T., & Manning, C. E. (2019). Subducting carbon. Nature ,574(7778), 343–352. Pletsch, T. (2001). Palaeoe nviron mental i mplications of palygorskite clays in Eoce ne deep ‐water sedi ments fro m the wester n central Atla ntic. Geological Society of London, Special Publication ,183(1), 307–316. Poli, S. (2015). Carbon mobilized at shallo w depths in subduction zones by carbonatitic liquids. Nature Geoscience ,8(8), 633. Presti, M., & Mic halopoulos, P. (2008). Esti mati ng t he co ntributio n of t he aut hige nic mi neral co mpo ne nt to t he lo ng ‐ter m reactive silica accu mulation on the western shelf of the Mississippi River Delta. Continental Shelf Research ,28(6), 823–838. Q uirk, J., Leake, J. R., Jo h nso n, D. A., Taylor, L. L., Sacco ne, L., & Beerli ng, D. J. (2015). Co nstrai ni ng t he role of early la nd pla nts i n Palaeozoic weathering and global cooling. Proceedings of the Royal Society B ,282. Ra h ma n, S., Aller, R., & Coc hra n, J. (2016). Cos moge nic 32Si as a tracer of bioge nic silica b urial a nd diage nesis: Major deltaic si nks i n t he sili c a c y cl e. Geophysical Research Letters ,43, 7124–7132. https://doi.org/10.1002/2016 GL069929 Ra h ma n, S., Aller, R., & Coc hra n, J. (2017). T he missi ng silica si nk: Revisiti ng t he mari ne sedi me ntary Si cycle usi ng cos moge nic 32Si. Global Biogeoche mical Cycles ,31, 1559–1578. https://doi.org/10.1002/2017 G B005746 Ras musse n, B., Krape ž, B., Mu hli ng, J. R., & Suvorova, A. (2015). Precipitatio n of iro n silicate na noparticles i n early Preca mbria n ocea ns marks Earth's first iron age. Geology ,43(4), 303–306.

ISS O N ET AL. 2 6 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

Ras m usse n, B., Meier, D. B., Krape ž, B., & M u hli ng, J. R. (2013). Iro n silicate microgra n ules as prec ursor sedi me nts to 2.5‐billio n‐year‐old banded iron for mations. Geology ,41(4), 435–438. Ras musse n, B., Mu hli ng, J. R., Suvorova, A., & Krape ž, B. (2017). Gree nalite precipitatio n li nked to t he depositio n of ba nded iro n for ma- tions do wnslope fro m a late Archean carbonate platfor m. Preca mbrian Research,290, 49–62. Reeburgh, W., 1993, The role of methylotrophy in the global methane budget: Microbial gro wth on C ‐1 co mpounds, p. 1‐14. Ridg well, A. (2005). A Mid Mesozoic Revolution in the regulation of ocean che mistry. Marine Geology ,217(3), 339–357. Ristvet, B. L., 1978, Reverse weathering reactions within recent nearshore marine sedi ments, Kaneohe Bay, Oahu [Ph. D.: North western U ni v. Ronov, A., Yaroshevsky, A., & Migdisov, A. (1990). Che mical co mposition of the Earth's crust and geoche mical balance of main ele ments . Mosco w: Nauka, Science Pub. House. (in Russian) Rosas, J. C., & Kore naga, J. (2018). Rapid crustal gro wt h a nd ef fi cie nt crustal recycli ng i n t he early Eart h: I mplicatio ns for Hadea n a nd Archean geodyna mics. Earth and Planetary Science Letters ,494, 42–49. R uge nstei n, J. K. C., Ibarra, D. E., & vo n Bla ncke nb urg, F. (2019). Neoge ne cooli ng drive n by la nd s urface reactivity rat her t ha n i ncreased weathering fluxes. Nature ,571(7763), 99. Saal, A. E., Hauri, E. H., Lang muir, C. H., & Perfi t, M. R. (2002). Vapour undersaturation in pri mitive mid‐ocean‐ridge basalt and the volatile content of Earth's upper mantle. Nature ,419(6906), 451. Sagan, C., & Mullen, G. (1972). Earth and Mars: Evolution of at mospheres and surface te mperatures. Science,177(4043), 52–56. Sca mbelluri, M., Bebo ut, G. E., Bel monte, D., Gilio, M., Ca mpo menosi, N., Collins, N., & Crispini, L. (2016). Carbonation of subduction ‐ zone serpentinite (high‐pressure ophicarbo nate; Ligurian Western Alps) and i mplications for the deep carbon cycling. Earth and Planetary Science Letters ,441, 155–166. Sc holz, F., He nse n, C., Sc h midt, M., & Geerse n, J. (2013). S ub mari ne weat heri ng of silicate mi nerals a nd t he exte nt of pore water fres he ni ng at active continental margins. Geochi mica et Cos mochi mica Acta ,100, 200–216. Seiffert, F., Ba ndo w, N., Bouc hez, J., Vo n Bla ncke nburg, F., & Gorbus hi na, A. (2014). Microbial colo nizatio n of bare rocks: Laboratory bi o fi l m enhances mineral weathering. Procedia Earth and Planetary Science,10, 123–129. Siever, R. (1992). The silica cycle in the Preca mbrian. Geochi mica et Cos mochi mica Acta ,56(8), 3265–3272. Sillén, L. G. (1961). The physical che mistry of sea water. Oceanography ,67, 549–581. Sillén, L. G. (1967). The ocean as a che mical syste m. Science,156(3779), 1189–1197. Skelton, A. (2011). Flux rates for water and carbon during greenschist facies meta morphis m. Geology ,39(1), 43–46. Solo mon, E. A., Kastner, M., Jannasch, H., Robertson, G., & Weinstein, Y. (2008). Dyna mic fl uid fl o w and che mical fl uxes associated with a seafl oor gas hydrate deposit on the northern Gulf of Mexico slope. Earth and Planetary Science Letters ,270(1‐2), 95–105. Spivack, A. J., & Staudigel, H. (1994). Lo w‐te mperature alteratio n of t he upper ocea nic crust a nd t he alkali nity budget of sea water. Che mical Geology ,115(3‐4), 239–247. Stallard, R., & Ed mond, J. (1981). Geoche mistry of the A mazon: 1. Precipitation che mistry and the marine contrib ution to the dissolved load at the ti me of peak discharge. Journal of Geophysical Research,86( C10), 9844–9858. Stallard, R., & Ed mond, J. (1983). Geoche mistry of the A mazon: 2. The infl uence of geology and weathering environ ment on the dissolved load.Journal of Geophysical Research,88( C14), 9671–9688. Stallard, R. F. (1985). River che mistry, geology, geo morphology, and soils in the A mazon and Orinoco basi ns. In The che mistry of weath- eri ng, (pp. 293–316). Dordrec ht: Spri nger. Staudigel, H., & Hart, S. (1985). Dating of ocean crust hydrother mal alteration‐strontiu m isotope ratios fro m Hole‐504 B carbonates and a rei nterpretatio n of Sr isotope data fro m Deep‐Sea Drilli ng Project Sites 105, 332, 417, a nd 418. I nitial Reports of the Deep Sea Drilli ng Pr oject , 8 3 ( A P R), 2 9 7 – 3 0 3. Staudigel, H., Hart, S. R., Sc h mi ncke, H.‐U., & S mit h, B. M. (1989). Cretaceous ocea n crust at DS DP Sites 417 a nd 418: Carbo n uptake fro m weathering versus loss by mag matic outgassing. Geochi mica et Cos mochi mica Acta ,53(11), 3091–3094. Stein, C. A., & Stein, S. (1994). Constraints on hydrother mal heat fl ux through the oceanic lithosphere fro m global heat fl o w. Journal of Geophysical Research ,99( B2), 3081–3095. Ste wart, B. T., Sa ntos, I. R., Tait, D. R., Mackli n, P. A., & Ma her, D. T. (2015). Sub mari ne grou nd water disc harge a nd associated fl uxes of alkalinity and dissolved carbon into Moreton Bay ( Australia) esti mated via radiu m isotopes. Marine Che mistry ,174, 1–12. Ste wart, E., & Ague, J. J. (2018). Infiltration‐driven meta morphis m, Ne w England, US A: Regional C O 2 fl uxes and i mplications for Devonian cli mate and extinctions. Earth and Planetary Science Letters ,489, 123–134. Ste wart, E., Ag ue, J. J., Ferry, J. M., Sc hiffries, C. M., Tao, R. ‐B., Isso n, T. T., a nd Pla navs ky, N. J., 2019, Carbo natio n a nd decarbo natio n reactio ns: I mplicatio ns for pla netary habitability, Mi neralogical Society of A merica. S uc het, P.‐A., Probst, J. L., & L ud wig, W. (2003). World wide distrib utio n of co nti ne ntal rock lit hology: I mplicatio ns for t he at mosp heric/soil C O2 uptake by co nti ne ntal weat heri ng a nd alkali nity river tra nsport to t he ocea ns. Global Biogeoche mical Cycles ,17(2), 1038. https://doi.org/10.1029/2002 G B001891 Sun, X., & Turchyn, A. V. (2014). Signifi cant contribution of authigenic carbonate to marine carbon burial. Nature Geoscience ,7(3), 201. Szy mczycha, B., & Pe mpko wiak, J. (2015). The role of sub marine ground water discharge as material source to the Baltic Sea . Dordrecht: S pri n g er. Tajika, E., & Matsui, T. (1992). Evolution of terrestrial proto ‐C O2 at mosphere coupled with ther mal history of the earth. Earth and Planetary Science Letters ,113(1‐2), 251–266. Ta niguc hi, M. (2002). Tidal effects o n sub mari ne grou nd water disc harge i nto t he ocea n. Geophysical Research Letters ,29(12), 1561. https:// doi.org/10.1029/2002 GL014987 Ta niguc hi, M., Bur nett, W. C., Cable, J. E., & Tur ner, J. V. (2002). I nvestigatio n of sub mari ne grou nd water disc harge. Hydrological Processes ,16(11), 2115–2129. https://doi.org/10.1002/hyp.1145 Taniguchi, M., Burnett, W. C., Dulaiova, H., Kontar, E. A., Povinec, P. P., & Moore, W. S. (2006). Sub marine grou nd water discharge m e as ur e d by seepage meters i n Sicilia n coastal waters. Co ntinental Shelf Research ,26(7), 835–842. Tatzel, M., vo n Bla ncke nb urg, F., Oelze, M., Sc h uessler, J. A., & Bo hr ma n n, G. (2015). T he silico n isotope record of early silica diage nesis. Earth and Planetary Science Letters ,428, 293–303. Torres, M. A., West, A. J., & Li, G. (2014). S ulp hide oxidatio n a nd carbo nate dissol utio n as a so urce of C O2 over geological ti mescales. Nature ,507(7492), 346–349. Tosca, N. J., Gugge n hei m, S., & Pufa hl, P. K. (2016). A n aut hige nic origi n for Preca mbria n gree nalite: I mplicatio ns for iro n for matio n a nd t he c he mistry of a ncie nt sea water. Geological Society of A merica Bulletin ,128(3‐4), 511–530.

ISS O N ET AL. 2 7 of 2 8 Global Biogeoche mical Cycles 10.1029/2018 GB006061

Tosca, N. J., Macdo nald, F. A., Stra uss, J. V., Jo h nsto n, D. T., & K noll, A. H. (2011). Sedi me ntary talc i n Neoproterozoic carbo nate s uc- cessions. Earth and Planetary Science Letters ,306(1), 11–22. Tréguer, P. J., & De La Rocha, C. L. (2013). The world ocea n silica cycle. Annual Revie w of Marine Science ,5, 477–501. Urey, H. C. (1952). O n t he early c he mical history of t he Eart h a nd t he origi n of life. Proceedi ngs of the Natio nal Acade my of Scie nces ,38(4), 3 5 1 – 3 6 3. Va n Der Meer, D. G., Zeebe, R. E., va n Hi nsberge n, D. J., Sl uijs, A., Spak ma n, W., & Torsvik, T. H. (2014). Plate tecto nic co ntrols o n at mospheric C O2 levels since the Triassic. Proceedings of the National Acade my of Sciences ,111(12), 4380–4385. Voice, P. J., Ko wale wski, M., & Eriksso n, K. A. (2011). Qua ntifyi ng t he ti mi ng a nd rate of crustal evolutio n: Global co mpilatio n of radio- metrically dated detrital zirco n grai ns. The Journal of Geology ,119(2), 109–126. Von Strand mann, P. A. P., Jenkyns, H. C., & Wood fine, R. G. (2013). Lithiu m isotope evidence for enhanced weathering during Oceanic Anoxic Event 2. Nature Geoscience ,6(8), 668. Walker, J. C., Hays, P., & Kasti ng, J. F. (1981). A negative feedback mec ha nis m for t he lo ng ‐ter m stabilizatio n of Eart h's surface te m- perature. Journal of Geophysical Research,86( C10), 9776–9782. Wall mann, K., and Aloisi, G., 2012, The global carbon cycle: Geological processes Funda mentals of Geobiology 20 ‐35. Wall ma n n, K., Aloisi, G., Haeckel, M., Tis hc he nko, P., Pavlova, G., Grei nert, J., et al. (2008). Silicate weat heri ng i n a noxic mari ne sedi- ments. Geochi mica et Cos mochi mica Acta ,72(12), 2895–2918. Wall ma n n, K., Pi nero, E., Bur wicz, E., Haeckel, M., He nse n, C., Dale, A., & Ruepke, L. (2012). T he global i nve ntory of met ha ne hydrate i n marine sedi ments: A theoretical approach. Energies ,5(7), 2449–2498. Wilde, S. A., Valley, J. W., Peck, W. H., & Gra ha m, C. M. (2001). Evide nce fro m detrital zirco ns for t he existe nce of co nti ne ntal cr ust a nd oceans on the Earth 4.4 Gyr ago. Nature ,409(6817), 175. Winnick, M. J., & Maher, K. (2018). Relatio nships bet ween C O2, ther modyna mic li mits on silicate weathering, and the strength of the silicate weathering feedback. Earth and Planetary Science Letters ,485, 111–120. Wort ma n n, U. G., & Payta n, A. (2012). Rapid variability of sea water c he mistry over t he past 130 millio n years. Science,337(6092), 334–336. Z ha ng, S., Ague, J. J., & Brovaro ne, A. V. (2018). Degassi ng of orga nic carbo n duri ng regio nal meta morp his m of pelites, Wepa waug Sc hist, Connecticut, US A. Che mical Geology , 4 9 0 , 30 – 4 4. Z ha ng, S., & Pla navsky, N. (2019a). Revisiti ng grou nd water fl uxes to t he ocea n wit h i mplicatio ns for t he carbo n cycle. Geology ,48(1), 67–71. Z ha ng, S., & Pla navsky, N. J. (2019b). T he silicate weat heri ng feedback i n t he co ntext of op hiolite e mplace me nt: I nsig hts fro m a n i nverse model of global weathering proxies. A merican Journal of Science ,319(2), 75–104.

ISS O N ET AL. 2 8 of 2 8