Article | Open | Published:

# Cenozoic global cooling and increased seawater Mg/Ca via reduced reverse weathering

## Abstract

Authigenic clay minerals formed on or in the seafloor occur in every type of marine sediment. They are recognized to be a major sink of many elements in the ocean but are difficult to study directly due to dilution by detrital clay minerals. The extremely low dust fluxes and marine sedimentation rates in the South Pacific Gyre (SPG) provide a unique opportunity to examine relatively undiluted authigenic clay. Here, using Mg isotopes and element concentrations combined with multivariate statistical modeling, we fingerprint and quantify the abundance of authigenic clay within SPG sediment. Key reactants include volcanic ash (source of reactive aluminium) and reactive biogenic silica on or shallowly buried within the seafloor. Our results, together with previous studies, suggest that global reorganizations of biogenic silica burial over the Cenozoic reduced marine authigenic clay formation, contributing to the rise in seawater Mg/Ca and decline in atmospheric CO2 over the past 50 million years.

## Introduction

The formation of marine authigenic clays is one of the most enigmatic and least understood of the major processes involved in the global geochemical cycling of elements and alkalinity in the ocean1,2,3,4. Early work on chemical mass balance in the oceans hypothesized that authigenic clay formation via a reverse weathering reaction was one of the principle sinks of Si, Mg, K, Na, and alkalinity from seawater5, 6.

$${\rm{Si}}{{\rm{O}}_{{\rm{2(s)}}}} + {\rm{Al}}{\left( {{\rm{O}}{{\rm{H}}_{\rm{3}}}} \right)_{{\rm{(s)}}}} + \left( {{\rm{M}}{{\rm{g}}^{{\rm{2 + }}}},{{\rm{K}}^{\rm{ + }}},{\rm{N}}{{\rm{a}}^{\rm{ + }}}} \right) + {\rm{HCO}}_{3({\rm{aq}})}^ - \\ \to {\rm{new}}\,{\rm{clay}}\,{\rm{mineral + C}}{{\rm{O}}}_{2({\rm{aq}})} + {{\rm{H}}_{2}}{\rm{O}}$$
(1)

Although the discoveries of high and low-temperature hydrothermal systems at mid-ocean ridges identified a new sink for elements in the ocean7, 8, more recent studies indicate that the formation of authigenic clays in both shallow and deep-sea sediments of diverse lithologies is indeed widespread and likely quantitatively important for the geochemical budgets of cations and alkalinity in seawater9,10,11,12,13,14,15,16,17,18.

In this study, we fingerprint and quantify the abundance of authigenic clays within marine sediment from the largest oceanic province in the world, the South Pacific Gyre (SPG). Drilled during Integrated Ocean Drilling Program (IODP) Expedition 329, the pelagic sediment we examine is completely oxic, homogenous, red−brown, zeolitic, and metalliferous, and also includes abundant authigenic clay minerals19, 20. Far from land and in a region of low oceanic primary productivity, sedimentation rates in the SPG are among the slowest in the world ( ~ 0.1–1 m/Myr or μm/year)19, 21. Thus, the authigenic clays within the sediment are relatively undiluted by dust and biogenic deposition22, making the SPG an optimal location to study the geochemical signatures of and processes driving marine authigenic clay formation.

Our results support previous observations of the components and conditions required to form marine authigenic clays9,10,11,12,13,14,15,16,17,18 and highlight the importance of volcanic ash and biogenic Si as key reactants in the deep sea. In most ocean regions, the formation of authigenic clay is limited by the presence of reactive silica and we use our results to examine how the geographic reorganization of Si accumulation on the seafloor changed the global production of authigenic clay through time. We propose that shifts in biogenic silica burial decreased authigenic clay formation and could be responsible, in large part, for the increase in seawater Mg/Ca ratio observed over the Cenozoic. Considering the involvement and associated feedback loops of the global carbon cycle, reduced reverse weathering may also have contributed significantly to the decline in atmospheric CO2 and thus global cooling over the past 50 million years.

## Results

### The geochemical fingerprint of marine authigenic clay

We identify authigenic clay, which is broadly aluminosilicate in composition, in sediment from the SPG using two independent approaches. First, we use geochemical partitioning techniques and construct multivariate statistical mixing models based on a data set including major and trace element concentrations in 138 bulk pelagic sediment samples from IODP Sites U1365, U1366, U1367, U1369, U1370, and U137122. We used Q-mode factor analysis of the data set to identify the authigenic end-members and constrained least square multiple linear regression models to quantify their abundance (i.e., mass fraction) within the bulk sediment (Supplementary Note 1)22,23,24. Including nine major elements (Si, Al, Ti, Fe, Mn, Ca, Mg, K, and P) and a few trace elements (Cr, Rb, Cs), both statistical techniques quantitatively identify that the sediments are a mixture of six end-members, namely, dust, Fe/Mn-oxyhydroxides, biogenic Si, apatite, and two compositionally different altered volcanic ashes (Fig. 1a, b, and Supplementary Note 1, Supplementary Figs. 13, Supplementary Table 1). The two altered ashes were sampled from discrete ash layers at Sites U1366 and U1370 and their compositions are enriched in Mg and K, respectively.

Second, we measure the magnesium isotope ratio (δ26Mg) of the bulk sediment at Site U1366. Previous studies of both terrestrial25 and marine2 environments indicate that the incorporation of Mg into authigenic clays involves the preferential uptake of the heavier Mg isotope (26Mg) and that the Mg isotopic fractionation associated with clay formation ranges from 0 to + 1.25 per mil (‰) relative to the solution. As a result, authigenic marine clays should be enriched in 26Mg compared to seawater (δ26Mg = − 0.82‰) and bulk-silicate earth (δ26Mg = − 0.29‰). Measured δ26Mg values of the Site U1366 bulk sediment ranges from − 0.68‰ to + 0.31‰ (Fig. 1c, Supplementary Table 1). The Mg-enriched altered ash layer used as an end-member in the bulk chemistry statistical models had an even higher δ26Mg signature of + 0.52‰, which is one of the highest δ26Mg values ever recorded in solid-phase marine sediment and is 1.3‰ heavier than seawater (Supplementary Fig. 4). Other than the ash layer, the bulk sediments that yielded higher modeled mass fractions of the Mg-enriched altered ash in the statistical models had consistently heavier δ26Mg values (Fig. 2). Agreement between the two geochemical techniques is strong evidence that the Mg-enriched altered ash represents a clay component that has incorporated Mg directly from seawater into its mineral structure.

### The global marine authigenic clay sink

Quantifying the global sink of seawater elements in authigenic marine clays requires an understanding of both the kinetics and the ingredients that promote authigenic clay formation. Two lines of evidence suggest that the formation of authigenic clays in the SPG was geologically rapid and occurred at or near the seafloor, rather than deeply buried. First, porewater profiles of Mg concentrations and δ26Mg are constant with depth at each site19 (Supplementary Fig. 5), indicating that clays are not actively removing or releasing Mg into deeply buried porewaters. Second, the linear relationship between the modeled mass fraction of Mg-enriched altered ash and their δ26Mg values (Fig. 2) indicates mixing of two end-members with distinct δ26Mg compositions. If clay authigenesis had occurred deeper in the sediment column, Rayleigh-type distillation of Mg in the pore-fluid would have resulted in variable δ26Mg values for the authigenic clay end-member2, 26.

The often-described reverse weathering reaction that forms authigenic clay5, 9 requires three ingredients including a reactive Al source (“amorphous aluminosilicate”), a reactive Si source, and cations (e.g., Mg2+, K+, Fe2+,3+, Na+). Our isotopic and provenance results agree with this recipe for authigenic marine clays, although differences exist in the source of each reactant supplying the open ocean versus regions closer to continents, as discussed here.

The geochemical provenance models in our study suggest that volcanic ash is the primary precursor substrate to the authigenic clays forming in the deep-sea pelagic sediment in the SPG22. Volcanic ash is an excellent candidate to undergo alterations on or within the seafloor because it is deposited as very fine-grained amorphous glass susceptible to dissolution and is known to alter into smectites and zeolites, which have high cation exchange capacities27, 28. Studies of porewater profiles and authigenic minerals have hypothesized that altered volcanic ash is a solid phase in marine sediment consuming Mg, K, Li, and other cations from porewaters13, 27,28,29,30. Volcanic ash is also ubiquitous in bulk marine sediment31 in globally significant abundances32, and is deposited as discrete layers and/or dispersed ash that has been mixed into bulk sediment from ash layers via bioturbation or the settling of airborne or subaqueous ash. In short, the abundance of volcanic ash in the deep sea is vastly greater than is accounted for in the visible discrete volcanic ash layers. Volcanic ash may also be present in shallower water sediment systems, but eolian dust or riverine sediment may be a more important source of degraded minerals closer to the continents.

At Site U1366, our model results indicate that the abundance of Mg-enriched altered ash is related to the accumulation rate of biogenic Si (Fig. 1). The accumulation of biogenic Si is an important source of reactive Si for authigenic clay formation, in addition to any Si that is also released during ash alteration14, 28. Closer to river deltas and continental shelves, dissolved reactive Si may be supplied directly by river or from biogenic siliceous deposition10, 33.

Seawater provides an ample supply of Mg2+ (and other cations such as K+ and Na+) to be consumed during reverse weathering reactions. There may also be local sources of Mg within the sediment from the dissolution of aluminosilicate minerals, but these are likely to be relatively minor in early authigenesis compared to contributions from seawater.

In summary, an intimate mixture of biogenic Si with abundant and ubiquitous volcanic ash on or shallowly buried beneath the seafloor appears to be the most conducive environment to forming authigenic clay in deep-sea pelagic sediment. With this recipe we proceed to estimate the rate and extent of authigenic clay formation and quantify its global importance as a sink for seawater Mg and alkalinity and how it may have changed over the Cenozoic.

### Global reorganization of Si burial limits clay formation

In most ocean regions, the availability of reactive Si limits the formation of authigenic clays9, 15. The exceptions are in the Southern Ocean, and a few other smaller regions, where rapid accumulation of siliceous material significantly increases the Si concentrations of the porewater in the shallowly buried seafloor and the supply of Al limits the formation of authigenic clays16. Thus, changes in the geographic distribution, accumulation, and mixing of these components control the extent to which authigenic clays form in sediments globally.

Reconstructions of the abundance and spatial patterns of silica deposition over the Cenozoic have revealed profound changes in the distribution and mode of silica burial34,35,36,37. At present, the accumulation of siliceous material is primarily focused in the Southern Ocean. During the Paleogene, and in particular during the Early Eocene Climatic Optimum (EECO), biogenic silica accumulation and its diagenetic products (e.g., porcellanites and chert)38 were more widely deposited throughout the oceans34,35,36,37. The global occurrence of chert reached a peak, despite overall open ocean biosiliceous productivity being relatively low36. The presence of metals within opal and siliceous diagenetic products is known to lower mineral solubility, and authigenic clay minerals are commonly found in intimate association16, 17, 39. Together these observations suggest a global silica cycle in the Paleogene that was considerably more favorable for the formation of authigenic marine clays.

In the later Eocene and Oligocene, biogenic Si accumulation shifted from being dispersed in the open ocean to being focused beneath more localized upwelling regimes, reducing the spatial extent of reactive Si available to the seafloor. This change is documented most notably in the Southern Ocean where formerly silica-poor sediments were replaced by the largest sink of biogenic silica on Earth37. The trigger for the shift may have been changes in ocean circulation from the opening or closing of ocean gateways40, and/or the concurrent evolutionary radiation and proliferation of diatoms, which require concentrated silicic acid to grow37. Regardless of the trigger, a large spatial shift in the accumulation of siliceous material through the Cenozoic is clearly evident in the sedimentary record.

A consequence of the redistribution of global biogenic silica burial over the Cenozoic is that two of the critical ingredients to form marine authigenic clays—reactive Al and reactive Si—became geographically isolated from one another. Given that marine authigenic clay formation appears to be limited by the availability of one or both of these ingredients9, 16, the un-mixing of these components also likely resulted in a decline in the global marine authigenic clay sink over the Cenozoic. A reduction in authigenic clay would decrease the Mg removed from seawater and may explain most of the change in global Mg budgets observed over the Cenozoic.

### Reduced authigenic clay formation increased seawater Mg/Ca

Reconstructions of seawater Mg/Ca from a wide range of geochemical proxies over the Cenozoic indicate that seawater Mg/Ca has undergone a 2 to 3-fold increase, driven by both an increase in seawater Mg and a decline in seawater Ca41, 42. Higgins and Schrag43 used reconstructions of the δ26Mg composition of seawater from bulk foraminifera and fossil corals over the Cenozoic to show that the increase in seawater Mg flux was  ~ 3 Tmol/year and likely related to changes in the aluminosilicate flux, rather than variations in carbonate deposition or carbonate weathering. That is, the  ~ 3 Tmol/year change was caused by an increase in Mg sources from silicate weathering or a decline in Mg sinks in authigenic clays forming in marine sediments or low-temperature hydrothermal systems.

The mechanism favored by Higgins and Schrag43 was a decline in the Mg sink in marine clays due to cooler global temperatures slowing reaction kinetics. They speculated that the sink of Mg in authigenic clays might have a temperature dependence similar to off-axis low-temperature hydrothermal systems (i.e., basalt alterations from fluid flowing through the basement rock), but there are significant uncertainties in this comparison44,45,46,47,48,49. Here, we propose an alternative explanation that is independent of, but may complement, any temperature-dependent Mg removal in either system. The shift in the spatial distribution of biogenic silica accumulation in the ocean over the Cenozoic and an un-mixing of the ingredients required for the formation of marine authigenic clays caused a decline in the Mg sink in sedimentary authigenic marine clays (Fig. 3)

.

Our study provides strong empirical evidence that Mg in the SPG clays was taken up from seawater and allows us to produce first-order, quantitative constraints on the importance of authigenic clay formation as a sink for Mg. Given the enrichment of Mg in the altered ash component from this study, we extrapolate our results to estimate that typical deep-sea sediment removes ~ 0.02 Tmol of Mg from seawater per year (Supplementary Note 2). Although this present-day Mg sink in these pelagic environments is small, Si-enriched bulk sediment sampled from near the chert layer at Site U1365 indicate higher rates of Mg removal that would constitute a significant sink of seawater Mg if occurring on a global scale. For example, if covering 50−100% of the seafloor, this type of sedimentation would remove 0.4−0.8 Tmol of Mg from seawater per year (Supplementary Note 2). These estimates are  ~ 20−35% of the modern riverine influx of Mg from silicate weathering and are the same order of magnitude ( ~ Tmol) as the changes required by models to drive the observed increase in seawater Mg/Ca over the Cenozoic43.

The amount of Mg removed by authigenic clay formation is almost certainly even higher in more rapidly accumulating sediments such as river deltas, continental shelves and margins. For example, in the Amazon Delta and mobile mud belts that extend ~ 1600 kilometers away from the continent, ~ 90% of the Si deposition is transformed into authigenic marine clays during burial10, 33, 39. Because Si limits authigenic clay formation in these environments, enhanced deposition of biogenic Si during the Paleogene would have promoted the formation of additional authigenic marine clays. In summary, our calculations, combined with estimates for clay formation in shallow water environments10, suggest that marine authigenic clays remove Mg on a scale that could explain the change in seawater Mg/Ca over the Cenozoic.

### Cenozoic global cooling via decreased reverse weathering

The apparent decline in atmospheric CO2 and associated global cooling through the Cenozoic is commonly attributed to various mechanisms including an increase in the rate of global silicate weathering50, 51, changes in weatherability52, 53, a decline in volcanic outgassing54, 55, or some combination of the three. Here, we propose a fourth possibility—a decrease in reverse weathering in marine sediments. The mechanism is possible because reverse weathering is predicted to remove alkalinity from seawater via the reaction described in Eqn (1) with the total magnitude of the alkalinity sink depending on the relative importance of the reactant cations in the authigenic clay (weighted by charge).

A reduction in the amount of authigenic clay formation would reduce the alkalinity consumed by reverse weathering leading to an increase in the ocean’s alkalinity inventory (sources > sinks), an increase in seawater pH, and a decline in atmospheric CO2 (Fig. 3). The decrease in atmospheric CO2 would, in turn, result in global cooling and lowered rates of continental silicate and/or seafloor weathering until the point at which it balances the reduced alkalinity sink in authigenic marine clays (sources ≈ sinks). This mechanism is capable of changing rates of continental silicate weathering without changing either the rate of volcanic outgassing54, 55 or the rate constant for continental weatherability52, 53 and as a result represents a unique way of modulating the global carbon and alkalinity cycles and Earth’s climate on geologic timescales.

Assuming the  ~ 3 Tmol/year change in Mg sink over the Cenozoic predicted from models of Mg/Ca ratios and δ26Mg composition of seawater is driven entirely by changes in reverse weathering, the corresponding change in alkalinity consumption (CO2 production) is 6 × 1012 alkalinity equivalents per year (or 3 Tmol C/year). That amount is ~ 25% of the present-day CO2 consumption from silicate weathering ( ~ 12 Tmol C/year)56 and is roughly similar to the budgets of Mackenzie and Garrels5 which estimated that reverse weathering reactions involving Mg removed  ~ 30% of the river influx of alkalinity. Additional reverse weathering reactions involving Na+ and K+ may increase the importance of this alkalinity sink. Even when considering only reverse weathering reactions associated with Mg (that is, neglecting the potential contributions of Na and K alterations), our first-order estimates indicate that this change would result in a doubling of atmospheric CO2 from common era (280 p.p.m. to 590 p.p.m.), assuming a global carbon cycle at steady-state and silicate weathering that depends on pCO2 (Supplementary Note 3). This estimate is on the low end but within the range of atmospheric CO2 levels in the mid to late Eocene57, suggesting that a decline in the marine authigenic clay sink may have played an underappreciated and significant role in facilitating the decline in atmospheric CO2 and associated global cooling over the Cenozoic.

## Methods

### Element concentrations

The data set of bulk sediment major, trace, and rare earth element concentrations used in this study is from Dunlea et al.21, 22, in which are described the analytical methods and age model in detail. The same acid digestions of samples that were prepared and analyzed for their trace and rare earth element compositions were also processed and analyzed for Mg isotopes at Princeton University, using techniques described below.

### Multivariate statistical modeling

The Q-mode factor analysis and constrained least squares multiple linear regressions were performed using MATLAB codes published in Pisias et al.23, and Dunlea and Murray24. Many iterations of the QFA and CLS analyses were run with different sample selections and element menus to check the robustness and stability of the results. Additional details of the statistical modeling and stability tests are discussed in Supplementary Note 1.

### Magnesium isotopes

Mg ions were separated with an automated ion chromatography (IC) Dionex ICS-5000 + IC system coupled with a Dionex AS-AP fraction collector. Samples were then injected with 0.2% HNO3 and passed through a Dionex CS-16 cation-exchange column using a methyl-sulfonic acid (MSA) eluent at a flow rate of 1 ml/min. The conductivity of each sample was measured and the sample was collected during the desired time window. Including column washing, elution time per sample ranges from 30 to 50 min, depending on the sample matrix. Blank levels yielded < 5 ng for Mg, or < 1% of the sample mass (500 ng for Mg). Following ion separation, samples were treated with concentrated HNO3, dried and re-diluted to 150–200 p.p.b. in 2% HNO3, and analyzed on the Thermo Scientific Neptune Plus multi-collector inductively coupled plasma mass spectrometer (MC-ICP-MS) at Princeton University for their Mg isotope composition.

Sample-standard-sample bracketing was used to correct for instrumental mass fractionation58 and measurements of inter-laboratory standards confirm the accuracy of these analytical methods. Mg isotopic compositions are reported using delta notation by which sample Mg isotope ratios are calculated relative to the standard DSM-3. The δ26Mg of the Cambridge-1 standard measured at  − 2.59 ± 0.06‰ and agrees well with measurements of other laboratories59.

### Data availability

The element concentration data used in this study are included in five tables in the supporting information files of Dunlea et al.22 and can be also downloaded from the EarthChem Library (http://get.iedadata.org/doi/100603, doi:10.1594/IEDA/100603)60. The δ26Mg values are reported in Supplementary Table 1 of this study. Any additional data may be obtained from AGD.

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## References

1. 1.

Mackenzie, F. T. & Kump, L. R. Reverse weathering, clay mineral formation, and oceanic element cycles. Science 270, 586–587 (1995).

2. 2.

Higgins, J. A. & Schrag, D. P. Constraining magnesium cycling in marine sediments using magnesium isotopes. Geochim. Cosmochim. Acta 74, 5039–5053 (2010).

3. 3.

Tréguer, P. J. & La Rocha De, C. L. The world ocean silica cycle. Annu. Rev. Mar. Sci. 5, 477–501 (2013).

4. 4.

Baldermann, A., Warr, L. N., Letofsky-Papst, I. & Mavromatis, V. Substantial iron sequestration during green-clay authigenesis in modern deep-sea sediments. Nat. Geosci. 8, 885–889 (2015).

5. 5.

Mackenzie, F. T. & Garrels, R. M. Chemical mass balance between rivers and oceans. Am. J. Sci. 264, 507–525 (1966).

6. 6.

Sillén, L. G. The ocean as a chemical system. Science 156, 1189–1197 (1967).

7. 7.

Von Damm, K. L., Edmond, J. M., Grant, B., Walden, B. & Weiss, R. F. Chemistry of submarine hydrothermal solutions at 21 N, East Pacific Rise. Geochim. Cosmochim. Acta 49, 2197–2220 (1985).

8. 8.

Elderfield, H. & Schultz, A. Mid-ocean ridge hydrothermal fluxes and the chemical composition of the ocean. Annu. Rev. Earth Planet. Sci. 24, 191–224 (1996).

9. 9.

Michalopoulos, P. & Aller, R. C. Rapid clay mineral formation in Amazon delta sediments: reverse weathering and oceanic elemental cycles. Science 270, 614–617 (1995).

10. 10.

Michalopoulos, P. & Aller, R. C. Early diagenesis of biogenic silica in the Amazon delta: alteration, authigenic clay formation, and storage. Geochim. Cosmochim. Acta 68, 1061–1085 (2004).

11. 11.

Yang, K. et al. Smectite, illite, and early diagenesis in South Pacific Gyre subseafloor sediment. Appl. Clay Sci. doi:10.1016/j.clay.2016.03.041 (2016).

12. 12.

Hurd, D. C. Interactions of biogenic opal, sediment and seawater in the Central Equatorial Pacific. Geochim. Cosmochim. Acta 37, 2257–2282 (1973).

13. 13.

Kastner, M. in The Oceanic Lithosphere, The Sea Vol. 7 (ed. Emiliani, C.) 915–980 (1981).

14. 14.

Cole, T. G. & Shaw, H. F. The nature and origin of authigenic smectites in some recent marine sediments. Clay Miner. 18, 239–252 (1983).

15. 15.

Cole, T. G. Composition, oxygen isotope geochemistry and origin of smectite in the metalliferous sediments of the Bauer Deep, southeast Pacific. Geochim. Cosmochim. Acta 49, 221–235 (1985).

16. 16.

Van Cappellen, P. & Qiu, L. Biogenic silica dissolution in sediments of the Southern Ocean. I. Solubility. Deep Sea Res. 44, 1109–1128 (1997).

17. 17.

Loucaides, S. et al. Seawater-mediated interactions between diatomaceous silica and terrigenous sediments: results from long-term incubation experiments. Chem. Geol. 270, 68–79 (2010).

18. 18.

Cuadros, J., Dekov, V. M., Arroyo, X. & Nieto, F. Smectite formation in submarine hydrothermal sediments: samples from the HMS challenger expedition (1872–1876). Clays Clay Miner. 59, 147–164 (2011).

19. 19.

D’Hondt, S., Inagaki, F. & Alvarez Zarikian, C. A. Expedition 329 Scientists. Expedition 329 Reports. Proc. IODP 329, doi:10.2204/iodp.proc.329.2011 (2011).

20. 20.

D’Hondt, S. et al. Presence of oxygen and aerobic communities from sea floor to basement in deep-sea sediments. Nat. Geosci. 8, 299–304 (2015).

21. 21.

Dunlea, A. G. et al. Cobalt-based age models of pelagic clay in the South Pacific Gyre. Geochem. Geophy. Geosyst. 16, 2694–2710 (2015).

22. 22.

Dunlea, A. G. et al. Dust, volcanic ash, and the evolution of the South Pacific Gyre through the Cenozoic. Paleoceanography 30, 1078–1099 (2015).

23. 23.

Pisias, N. G., Murray, R. W. & Scudder, R. P. Multivariate statistical analysis and partitioning of sedimentary geochemical data sets: general principles and specific MATLAB scripts. Geochem. Geophy. Geosyst. 5, 4015–4020 (2013).

24. 24.

Dunlea, A. G. & Murray, R. W. Optimization of end-members used in multiple linear regression geochemical mixing models. Geochem. Geophy. Geosyst. 16, 4021–4027 (2015).

25. 25.

Wimpenny, J., Colla, C. A., Yin, Q.-Z., Rustad, J. R. & Casey, W. H. Investigating the behaviour of Mg isotopes during the formation of clay minerals. Geochim. Cosmochim. Acta 128, 178–194 (2014).

26. 26.

Blättler, C. L., Miller, N. R. & Higgins, J. A. Mg and Ca isotope signatures of authigenic dolomite in siliceous deep-sea sediments. Earth Planet. Sci. Lett. 419, 32–42 (2015).

27. 27.

Hodder, A. P. W., Naish, T. R. & Nelson, C. S. A two-stage model for the formation of smectite from detrital volcanic glass under shallow-marine conditions. Mar. Geol. 109, 279–285 (1993).

28. 28.

Schacht, U., Wallmann, K., Kutterolf, S. & Schmidt, M. Volcanogenic sediment–seawater interactions and the geochemistry of pore waters. Chem. Geol. 249, 321–338 (2008).

29. 29.

Gieskes, J. M. & Lawrence, J. R. Alteration of volcanic matter in deep sea sediments: evidence from the chemical composition of interstitial waters from deep sea drilling cores. Geochim. Cosmochim. Acta 45, 1687–1703 (1981).

30. 30.

Martin, J. B., Kastner, M. & Elderfield, H. Lithium: sources in pore fluids of Peru slope sediments and implications for oceanic fluxes. Mar. Geol. 102, 281–292 (1991).

31. 31.

Scudder, R. P. et al. Geochemical approaches to the quantification of dispersed volcanic ash in marine sediment. Prog. Earth Planet. Sci. 3, doi:10.1186/s40645-015-0077-y (2016).

32. 32.

Olgun, N. et al. Surface ocean iron fertilization: The role of airborne volcanic ash from subduction zone and hot spot volcanoes and related iron fluxes into the Pacific Ocean. Global Biogeochem. Cycle 25, GB4001 (2011).

33. 33.

Rahman, S., Aller, R. C. & Cochran, J. K. Cosmogenic 32Si as a tracer of biogenic silica burial and diagenesis: major deltaic sinks in the silica cycle. Geophys. Res. Lett. 43, 7124–7132 (2016).

34. 34.

Keller, G. & Barron, J. Paleoceanographic implications of Miocene deep-sea hiatuses. Geol. Soc. Am. Bull. 94, 590–613 (1983).

35. 35.

Moore, T. Jr et al. Paleogene tropical Pacific: clues to circulation, productivity, and plate motion. Paleoceanography 19, PA3013 (2004).

36. 36.

Muttoni, G. & Kent, D. V. Widespread formation of cherts during the early Eocene climate optimum. Palaeogeogr. Palaeoclimatol. Palaeoecol. 253, 348–362 (2007).

37. 37.

Renaudie, J. Quantifying the Cenozoic marine diatom deposition history: links to the C and Si cycles. Biogeoscience 13, 6003–6014 (2016).

38. 38.

Murray, R. W. Chemical-criteria to identify the depositional environment of chert - general-principles and applications. Sediment. Geol. 90, 213–232 (1994).

39. 39.

Michalopoulos, P., Aller, R. C. & Reeder, R. J. Conversion of diatoms to clays during early diagenesis in tropical, continental shelf muds. Geology 28, 1095–1098 (2000).

40. 40.

Egan, K. E., Rickaby, R. E. M., Hendry, K. R. & Halliday, A. N. Opening the gateways for diatoms primes Earth for Antarctic glaciation. Earth Planet. Sci. Lett. 375, 34–43 (2013).

41. 41.

Coggon, R. M., Teagle, D., Smith-Duque, C. E., Alt, J. C. & Cooper, M. J. Reconstructing past seawater Mg/Ca and Sr/Ca from mid-ocean ridge flank calcium carbonate veins. Science 327, 1114–1117 (2010).

42. 42.

Higgins, J. A. & Schrag, D. P. Records of Neogene seawater chemistry and diagenesis in deep-sea carbonate sediments and pore fluids. Earth Planet. Sci. Lett. 357–358, 386–396 (2012).

43. 43.

Higgins, J. A. & Schrag, D. P. The Mg isotopic composition of Cenozoic seawater - evidence for a link between Mg-clays, seawater Mg/Ca, and climate. Earth Planet. Sci. Lett. 416, 73–81 (2015).

44. 44.

Coogan, L. A., Gillis, K. M., Pope, M. & Spence, J. The role of low-temperature (off-axis) alteration of the oceanic crust in the global Li-cycle: insights from the Troodos ophiolite. Geochim. Cosmochim. Acta 203, 201–215 (2017).

45. 45.

Coogan, L. A. & Gillis, K. M. Evidence that low-temperature oceanic hydrothermal systems play an important role in the silicate-carbonate weathering cycle and long-term climate regulation. Geochem. Geophy. Geosyst. 14, 1771–1786 (2013).

46. 46.

Gillis, K. M. & Coogan, L. A. Secular variation in carbon uptake into the ocean crust. Earth Planet. Sci. Lett. 302, 385–392 (2011).

47. 47.

Coogan, L. A. & Dosso, S. E. Alteration of ocean crust provides a strong temperature dependent feedback on the geological carbon cycle and is a primary driver of the Sr-isotopic composition of seawater. Earth Planet. Sci. Lett. 415, 38–46 (2015).

48. 48.

Fisher, A. T. & Wheat, C. G. Seamounts as conduits for massive fluid, heat, and solute fluxes on ridge flanks. Oceanography 23, 74–87 (2010).

49. 49.

Wheat, C. G. & Fisher, A. T. Massive, low-temperature hydrothermal flow from a basaltic outcrop on 23 Ma seafloor of the Cocos Plate: chemical constraints and implications. Geochem. Geophys. Geosyst. 9, doi:10.1029/2008GC002136 (2008).

50. 50.

Misra, S. & Froelich, P. N. Lithium isotope history of Cenozoic seawater: changes in silicate weathering and reverse weathering. Science 335, 818–823 (2012).

51. 51.

Raymo, M. E. & Ruddiman, W. F. Tectonic forcing of late cenozoic climate. Nature 359, 117–122 (1992).

52. 52.

Kump, L. R., Brantley, S. L. & Arthur, M. A. Chemical weathering, atmospheric CO2, and climate. Annu. Rev. Earth Planet. Sci. 28, 611–667 (2000).

53. 53.

Berner, R. A. Inclusion of the Weathering of Volcanic Rocks in the GEOCARBSULF Model. Am. J. Sci. 306, 295–302 (2006).

54. 54.

Berner, R. A., Lasaga, A. C. & Garrels, R. M. The carbonate-silicate geochemical cycle and its effect on atmospheric carbon dioxide over the past 100 million years. Am. J. Sci. 283, 641–683 (1983).

55. 55.

Lee, C.-T. A., Thurner, S., Paterson, S. & Cao, W. The rise and fall of continental arcs: Interplays between magmatism, uplift, weathering, and climate. Earth Planet. Sci. Lett. 425, 105–119 (2015).

56. 56.

Gaillardet, J., Dupré, B., Louvat, P. & Allegre, C. J. Global silicate weathering and CO2 consumption rates deduced from the chemistry of large rivers. Chem. Geol. 159, 3–30 (1999).

57. 57.

Pagani, M., Zachos, J. C., Freeman, K. H., Tipple, B. & Bohaty, S. Marked decline in atmospheric carbon dioxide concentrations during the Paleogene. Science 309, 600–603 (2005).

58. 58.

Galy, A., Belshaw, N. S., Halicz, L. & O’Nions, R. K. High-precision measurement of magnesium isotopes by multiple-collector inductively coupled plasma mass spectrometry. Int. J. Mass Spectrom. 208, 89–98 (2001).

59. 59.

Galy, A. et al. Magnesium isotope heterogeneity of the isotopic standard SRM980 and new reference materials for magnesium-isotope-ratio measurements. J. Anal. At. Spectrom. 18, 1352–1356 (2003).

60. 60.

Dunlea, A. G. et al. Major, trace, REE concentrations of bulk sediment squeeze cake samples from 7 sites in the South Pacific Gyre (IODP Expedition 329). Interdisciplinary Earth Data Alliance (IEDA). doi:10.1594/IEDA/100603 (2016).

## Acknowledgements

We thank T. Ireland, J.W. Sparks, and E. Lundstrom for their analytical assistance and R.P. Scudder, C.H. Anderson, C.C. McKinley, and E.R. Estes for their many conversations. This research used samples and data provided by the Integrated Ocean Drilling Program (IODP). Funding for this research was provided by the U.S. National Science Foundation to R.W.M. (OCE1130531) and to J.A.H. (OCE1654571). USSSP post-cruise support was provided to R.W.M. as an Expedition 329 shipboard participant. Portions of this material are based upon work supported while R.W.M. was serving at the National Science Foundation.

## Author information

### Affiliations

1. #### Department of Geology and Geophysics, Woods Hole Oceanographic Institution, 266 Woods Hole Road, Woods Hole, MA, 02543, USA

• Ann G. Dunlea
2. #### Department of Earth and Environment, Boston University, 685 Commonwealth Ave., Boston, MA, 02215, USA

• Richard W. Murray
3. #### Department of Geosciences, Princeton University, Guyot Hall, Princeton, NJ, 08544, USA

• Danielle P. Santiago Ramos
•  & John A. Higgins

### Contributions

A.G.D. constructed the multivariate statistical models, assisted with the isotope analyses, developed the comparison between the techniques and implications, and led the writing of the manuscript. D.P.S.R. performed the isotopic analyses and contributed key ideas and revisions to the manuscript. J.A.H. and R.W.M. made essential and substantive intellectual contributions and revisions to the ideas presented in the manuscript as well as provided the funds, resources, and oversight to conduct the geochemical analyses. Additionally, R.W.M sailed on IODP Expedition 329 and lovingly collected the terrific samples used in this study.

### Competing interests

The authors declare no competing financial interests.

### Corresponding author

Correspondence to Ann G. Dunlea.

## Electronic supplementary material

### DOI

https://doi.org/10.1038/s41467-017-00853-5