Figures

top

Introduction

The fractionation of garnet from subduction zone magmas is hypothesised to play a critical role in several fundamental arc processes: delamination of garnet-rich, density unstable cumulates may be critical to the production and stabilisation of continental crust (Jagoutz and Behn, 2013

Jagoutz, O., Behn, M.D. (2013) Foundering of lower island-arc crust as an explanation for the origin of the continental Moho. Nature 504, 131–134. https://doi.org/10.1038/nature12758

), while fractionation of ferrous iron-rich garnet may contribute to the generation of oxidised, calc-alkaline arc magmas (Tang et al., 2018

Tang, M., Erdman, M., Eldridge, G., Lee, C.-T.A. (2018) The redox “filter” beneath magmatic orogens and the formation of continental crust. Science Advances 4, eaar4444. https://doi.org/10.1126/sciadv.aar4444

). These hypotheses are supported by observations from exhumed arc lower crustal sections and xenoliths, which commonly include garnet-rich lithologies (Ducea and Saleeby, 1996

Ducea, M.N., Saleeby, J.B. (1996) Buoyancy sources for a large, unrooted mountain range, the Sierra Nevada, California: Evidence from xenolith thermobarometry. Journal of Geophysical Research: Solid Earth 101, 8229–8244. https://doi.org/10.1029/95jb03452

; Jagoutz, 2010

Jagoutz, O. (2010) Construction of the granitoid crust of an island arc. Part II: A quantitative petrogenetic model. Contributions to Mineralogy and Petrology 160, 359–381. https://doi.org/10.1007/s00410-009-0482-6

), and by experimental studies that show that garnet is a stable phase in hydrous magmas at typical arc lower crustal conditions (Alonso-Perez et al., 2009

Alonso-Perez, R., Müntener, O., Ulmer, P. (2009) Igneous garnet and amphibole fractionation in the roots of island arcs: experimental constraints on andesitic liquids. Contributions to Mineralogy and Petrology 157, 541–558. https://doi.org/10.1007/s00410-008-0351-8

; Ulmer et al., 2018

Ulmer, P., Kaegi, R., Müntener, O. (2018) Experimentally derived intermediate to silica-rich arc magmas by fractional and equilibrium crystallization at 1.0 GPa: An evaluation of phase relationships, compositions, liquid lines of descent and oxygen fugacity. Journal of Petrology 59, 11–58. https://doi.org/10.1093/petrology/egy017

). However, primary garnet phenocrysts are rare in typical arc lavas, making the ubiquity of garnet fractionation at modern arcs difficult to assess.

As an alternative approach, many researchers have highlighted the distinctive heavy rare earth element (HREE)-enriched trace element signature of garnet and argued that garnet must play an important role in the petrogenesis of evolved magmas with complementary HREE depletions. Early applications of this approach focused on Archean tonalite-trondhjemite-granodiorite suites (TTGs) and modern dacites with elevated Sr/Y and La/Yb ratios and argued that these magmas represent partial melts of garnet-bearing subducted crust (Drummond and Defant, 1990

Drummond, M.S., Defant, M.J. (1990) A model for trondhjemite‐tonalite‐dacite genesis and crustal growth via slab melting: Archean to modern comparisons. Journal of Geophysical Research: Solid Earth 95, 21503–21521. https://doi.org/10.1029/jb095ib13p21503

). More recently, trace element ratios including Sr/Y, Dy/Yb, and La/Yb were proposed as proxies for crustal thickness (Chapman et al., 2015

Chapman, J.B., Ducea, M.N., DeCelles, P.G., Profeta, L. (2015) Tracking changes in crustal thickness during orogenic evolution with Sr/Y: An example from the North American Cordillera. Geology 43, 919–922. https://doi.org/10.1130/g36996.1

; Profeta et al., 2015

Profeta, L., Ducea, M.N., Chapman, J.B., Paterson, S.R., Gonzales, S.M.H., Kirsch, M., Petrescu, L., DeCelles, P.G. (2015) Quantifying crustal thickness over time in magmatic arcs. Scientific Reports 5, 17786. https://doi.org/10.1038/srep17786

), based on the interpretation that they reflect increased extents of differentiation at pressures high enough to stabilise garnet (e.g., Davidson et al., 2007

Davidson, J., Turner, S., Handley, H., Macpherson, C., Dosseto, A. (2007) Amphibole “sponge” in arc crust? Geology 35, 787–790. https://doi.org/10.1130/g23637a.1

). These two approaches use the same trace element ratios to infer distinct processes at different locations within subduction zone systems. Therefore, in the absence of additional observations, interpretations of these ratios alone are necessarily non-unique. In this paper, we introduce melt MnO/MgO ratios as an alternative indicator of garnet fractionation based on two findings: primary mantle melts have comparatively uniform MnO/MgO ratios; and the Mn-Mg partition coefficient between garnet (gt) and melt, defined as K_D^Mn-Mg = (Mn_gt/Mn_melt)/(Mg_gt/Mg_melt), is greater than 1, unique among typical early fractionating phases.

top

Variability of Subduction Zone Primary Melts

Proxies for garnet differentiation and crustal thickness using trace element ratios, including Sr/Y and La/Yb, have been developed primarily for application to relatively evolved lavas. The rationale for these proxies is that fractionation of garnet-rich cumulates or, equivalently, extraction of partial melts from garnet-rich residues, causes depletions in HREEs, resulting in differentiated melts with elevated trace element ratios. While garnet fractionation may generate this signal, inferring this process based on only the values of these ratios in differentiated magmas requires independent knowledge of the derivative magma’s parental melt composition. Without this information, it is not possible to attribute elevated trace element ratios in differentiated magmas to garnet fractionation rather than to inherited parental magma compositions.

A recent compilation of primary arc lava compositions (Schmidt and Jagoutz, 2017

Schmidt, M.W., Jagoutz, O. (2017) The global systematics of primitive arc melts. Geochemistry, Geophysics, Geosystems 18, 2817–2854. https://doi.org/10.1002/2016gc006699

) highlights this problem. These samples have compositions in equilibrium with typical mantle peridotite, and thus have necessarily undergone minimal differentiation within the crust. However, these lavas display a wide range in both Sr/Y and La/Yb ratios that show no correlation with crustal thickness (Fig. 1). In contrast, both ratios are strongly correlated with primary melt type, regardless of crustal thickness or upper plate type: tholeiitic magmas have uniformly low Sr/Y and La/Yb ratios, while calc-alkaline magmas span a wide range. This large range in primary arc magma incompatible trace element ratios reflects several factors, including variable contributions from the slab (e.g., Elliott, 2004

Elliott, T. (2004) Tracers of the slab. In: Eiler, J. (Ed.) Inside the Subduction Factory. Geophysical Monograph Series 138, American Geophysical Union, Washington D.C., 23–45. https://doi.org/10.1029/138gm03

) and degree of prior mantle depletion (e.g., Kelley et al., 2006

Kelley, K.A., Plank, T., Grove, T.L., Stolper, E.M., Newman, S., Hauri, E. (2006) Mantle melting as a function of water content beneath back-arc basins. Journal of Geophysical Research: Solid Earth 111, B09208. https://doi.org/10.1029/2005jb003732

).

Full size image

During differentiation, a range of processes beyond garnet fractionation can further modify these ratios. Plagioclase accumulation generates magmas with elevated Sr and Sr/Y ratios (Vukadinovic, 1993

Vukadinovic, D. (1993) Are Sr enrichments in arc basalts due to plagioclase accumulation? Geology 21, 611–614.   https://doi.org/10.1130/0091-7613(1993)021< 0611:aseiab> 2.3.co;2

), while amphibole fractionation can produce magmas with elevated La/Yb ratios (Davidson et al., 2007

Davidson, J., Turner, S., Handley, H., Macpherson, C., Dosseto, A. (2007) Amphibole “sponge” in arc crust? Geology 35, 787–790. https://doi.org/10.1130/g23637a.1

). These additional processes are much less sensitive to pressure compared to garnet fractionation, and instead are dominantly sensitive to magmatic H₂O contents. Given the range of primary melt compositions and these additional confounding processes, it is unlikely that the trace element ratios of any single suite of differentiated arc magmas provide meaningful estimates of crustal thickness.

top

MnO/MgO Ratios in Primary Arc Melts

In contrast to incompatible trace element ratios, the behaviour of most compatible elements in mantle melts is controlled by melting reactions in the mantle wedge. Typically, these elements and their ratios vary predictably in response to changes in mantle melting regime (e.g., Grove et al., 2013

Grove, T.L., Holbig, E.S., Barr, J.A., Till, C.B., Krawczynski, M.J. (2013) Melts of garnet lherzolite: experiments, models and comparison to melts of pyroxenite and carbonated lherzolite. Contributions to Mineralogy and Petrology 166, 887–910. https://doi.org/10.1007/s00410-013-0899-9

), while some canonical ratios, such as Mg# (molar Mg/(Mg + Fe)), display a relatively invariant range reflective of equilibrium with mantle olivine. In Figure 2a, we show that the MnO/MgO ratios of primary arc magmas are also tightly clustered and, based on a recent model of olivine-melt Mn/Mg partitioning (Blundy et al., 2020

Blundy, J., Melekhova, E., Ziberna, L., Humphreys, M.C.S., Cerantola, V., Brooker, R.A., McCammon, C.A., Pichavant, M., Ulmer, P. (2020) Effect of redox on Fe–Mg–Mn exchange between olivine and melt and an oxybarometer for basalts. Contributions to Mineralogy and Petrology 175, 1–32. https://doi.org/10.1007/s00410-020-01736-7

), are consistent with control by equilibration with mantle olivine (Fig. 2b). These primary melt compositions are inconsistent with garnet-present mantle melting, as this process produces melts with distinctively low MnO/MgO ratios (≤0.01; Walter, 1998

Walter, M.J. (1998) Melting of garnet peridotite and the origin of komatiite and depleted lithosphere. Journal of Petrology 39, 29–60. https://doi.org/10.1093/petroj/39.1.29

). This is consistent with observations that most garnet is exhausted at moderate extents of melting at typical sub-arc mantle conditions (≤11 wt. % melt at 3 GPa; Kushiro, 1996

Kushiro, I. (1996) Partial Melting of a fertile mantle peridotite at high pressures: An experimental study using aggregates of diamond. In: Basu, A., Hart, S. (Eds.) Earth Processes: Reading the Isotopic Code. Geophysical Monograph Series 95, American Geophysical Union, Washington D.C., 109–122. https://doi.org/10.1029/gm095p0109

), and with final mantle equilibration at pressures lower than the spinel-garnet transition (Perrin et al., 2016

Perrin, A., Goes, S., Prytulak, J., Davies, D.R., Wilson, C., Kramer, S. (2016) Reconciling mantle wedge thermal structure with arc lava thermobarometric determinations in oceanic subduction zones. Geochemistry, Geophysics, Geosystems 17, 4105–4127. https://doi.org/10.1002/2016gc006527

).

Full size image

These primary lavas show a wide range of incompatible trace element ratios indicative of variable slab contributions (Fig. 1). However, their MnO/MgO ratios are uncorrelated with these trace element ratios (Fig. S-1) and show negligible evidence for control by slab contributions: subducted sediments have elevated MnO/MgO ratios compared to the sub-arc mantle (Plank, 2014

Plank, T. (2014) 4.17 - The chemical composition of subducting sediments. In: Holland, H.D., Turekian, K.K. (Eds.) Treatise on Geochemistry. Second Edition, Elsevier, Amsterdam, 607–629. https://doi.org/10.1016/b978-0-08-095975-7.00319-3

), and this elevated ratio is inherited by slab-derived melts or fluids despite equilibrating with garnet-rich residue (Schmidt et al., 2004

Schmidt, M.W., Vielzeuf, D., Auzanneau, E. (2004) Melting and dissolution of subducting crust at high pressures: the key role of white mica. Earth and Planetary Science Letters 228, 65–84. https://doi.org/10.1016/j.epsl.2004.09.020

). The apparent contradiction between incompatible trace element ratios and the MnO/MgO ratios is resolved with a simple mass balance argument: unlike incompatible trace elements, slab derived fluids and melts typically contain an order of magnitude less MnO and MgO than the final melt in equilibrium with the mantle, and thus can exert only minimal control on the eventual MnO/MgO ratios of these melts. The small fraction of the primary melts with elevated MnO/MgO ratios (Fig. 2a) may reflect contributions from unusually MnO-rich slab sediments, although these melts may also have undergone limited fractionation of olivine ± clinopyroxene (see below).

top

Methods and Data Compilation

The previous section demonstrates that mantle-derived melts show a restricted range of MnO/MgO ratios that are reflective of equilibration with mantle olivine. To characterise the evolution of this ratio during crystallisation differentiation, we compiled Mn-Mg mineral-melt partitioning data from published experimental studies, focusing on studies that include garnet-bearing experiments. As MnO is commonly included in experimental bulk compositions and is a standard electron microprobe analyte, we were able to compile large datasets of Mn/Mg partition coefficients for all typical experimental phases. The full list of compiled experimental references is available in the Supplementary Information References.

Although commonly measured, MnO is frequently present at <0.1 wt. % concentrations in experimental melts and is only rarely an emphasis of experimental studies (cf. Balta et al., 2011

Balta, J.B., Asimow, P.D., Mosenfelder, J.L. (2011) Manganese partitioning during hydrous melting of peridotite. Geochimica et Cosmochimica Acta 75, 5819–5833. https://doi.org/10.1016/j.gca.2011.05.026

). Further, in experiments conducted at relatively low temperatures, garnets can be strongly zoned. Due to these analytical complications, propagated relative uncertainties on experimental Mn/Mg partition coefficients (K_D^Mn-Mg) in many experiments, and particularly at lower temperatures and pressures relevant to garnet fractionation, are >50 % (1 standard deviation). To address these limitations, we reanalysed garnet and melt compositions from two previous experimental studies relevant to garnet crystallisation at crustal conditions (Alonso-Perez et al., 2009

Alonso-Perez, R., Müntener, O., Ulmer, P. (2009) Igneous garnet and amphibole fractionation in the roots of island arcs: experimental constraints on andesitic liquids. Contributions to Mineralogy and Petrology 157, 541–558. https://doi.org/10.1007/s00410-008-0351-8

; Ulmer et al., 2018

Ulmer, P., Kaegi, R., Müntener, O. (2018) Experimentally derived intermediate to silica-rich arc magmas by fractional and equilibrium crystallization at 1.0 GPa: An evaluation of phase relationships, compositions, liquid lines of descent and oxygen fugacity. Journal of Petrology 59, 11–58. https://doi.org/10.1093/petrology/egy017

). We used laser ablation ICP-MS to remeasure melt MnO contents and made new electron microprobe measurements of garnet rim major element contents from 25 experiments. Detailed descriptions of analytical methods are available in the Supplementary Information, and the new analyses are presented in Table S-1. All compiled garnet partitioning data are shown in Figure 3d. As our primary focus here is on the role of garnet in subduction zone magmas, we subsequently exclude experiments conducted at pressures >5 GPa or those containing significant amounts of CO₂ or where the standard deviation uncertainty (1σ) on garnet K_D^Mn-Mg is ≥50 %.

Full size image

top

Results and Discussion

Our compiled partitioning data shows that garnet is unique among commonly fractionating minerals: most silicate phases are characterised by K_D^Mn-Mg consistently <1, and typically <0.5 (Fig. 3a–c), while garnet K_D^Mn-Mg shows considerably more variation but is consistently higher than other phases (≥1; Fig. 3d). In contrast to non-garnet silicates, Fe-Ti oxides also typically have K_D^Mn-Mg > 1 (Fig. S-2). However, these oxides incorporate much smaller amounts of MnO and MgO compared to garnet and are usually less abundant, and thus their fractionation has a comparatively smaller influence on melt compositions. To better constrain the effect of garnet fractionation, we modelled the garnet-melt K_D^Mn-Mg data using an Arrhenius type function (Fig. 3e). We explored models with additional parameters including melt and garnet compositions, and fO₂, but did not find that incorporating extra variables improved model performance. Additional experiments may reveal other variables that influence garnet K_D^Mn-Mg. However, given the robust model fit presented here, we expect other variables to be secondary to the effects of changing temperature and pressure.

The distinctive partitioning behaviour of garnet makes MnO/MgO ratios ideal for isolating the impact of garnet fractionation. Unlike MnO/FeO ratios, which were previously used to identify distinct mantle melt source regions (e.g., Balta et al., 2011

Balta, J.B., Asimow, P.D., Mosenfelder, J.L. (2011) Manganese partitioning during hydrous melting of peridotite. Geochimica et Cosmochimica Acta 75, 5819–5833. https://doi.org/10.1016/j.gca.2011.05.026

), the evolution of MnO/MgO ratios are insensitive to variations in magmatic fO₂: while Mn can occur in multiple valence states, the MnO-Mn₃O₄ fO₂ buffer is more than 4 log units above QFM, and thus Mn³⁺ is negligible in typical arc magmas (O’Neill and Pownceby, 1993

O’Neill, H.St.C., Pownceby, M.I. (1993) Thermodynamic data from redox reactions at high temperatures. II. The MnO–Mn₃O₄ oxygen buffer, and implications for the thermodynamic properties of MnO and Mn₃O₄. Contributions to Mineralogy and Petrology 114, 315–320. https://doi.org/10.1007/bf01046534

). In comparison, the partitioning of MnO from total iron will inevitably introduce fO₂ sensitivity due to the significant quantities of Fe³⁺ present at typical magmatic fO₂ conditions. Additionally, apart from fO₂ effects, MnO and FeO are much less strongly partitioned by most silicate phases, significantly complicating the ability to isolate the effects of garnet fractionation with MnO/FeO ratios.

Using our model for garnet K_D^Mn-Mg, we can estimate how melt MnO/MgO ratios evolve in response to fractionation of various phases. We find that at pressures <2 GPa, the garnet K_D^Mn-Mg is >1 for any magma below 1300 °C. At conditions more typical to arc lower crust, the garnet K_D^Mn-Mg increases to values >2, and thus pure garnet fractionation will cause melt MnO/MgO ratios to decrease. The effect of garnet fractionation is further amplified when comparing the effect of garnet fractionation in MnO/MgO to SiO₂. As garnet has much lower SiO₂ contents than typical arc melts, garnet fractionation will cause rapid increases in melt SiO₂ while hindering increases in MnO/MgO ratios, a distinctive trend in comparison to most other phases that crystallise from arc melts.

The impact of garnet fractionation is evident when examining the SiO₂ contents and MnO/MgO ratios of typical arc melts (Fig. 4). Clear differences are immediately apparent between relatively thin-crusted island arcs and continental arcs with thicker upper plates. Melts in both settings originate near the primitive arc compositions shown in Figure 2, but island arc magmas rapidly evolve to significantly higher MnO/MgO values, consistent with initial olivine + clinopyroxene dominated fractionation (Fig. 4a). Lavas from the westernmost Aleutian Arc and from Matthew and Hunter Volcanoes in Vanuatu are notable exceptions to the general island arc trend and are highlighted in Figure 4a. Erupted lavas at these locations are generally evolved and have very low MnO/MgO ratios, likely requiring the involvement of garnet in their petrogenesis. This is consistent with independent evidence that supports a slab-derived origin for these magmas (Yogodzinski et al., 2015

Yogodzinski, G.M., Brown, S.T., Kelemen, P.B., Vervoort, J.D., Portnyagin, M., Sims, K.W.W., Hoernle, K., Jicha, B.R., Werner, R. (2015) The role of subducted basalt in the source of island arc magmas: Evidence from seafloor lavas of the Western Aleutians. Journal of Petrology 56, 441–492. https://doi.org/10.1093/petrology/egv006

; McCarthy et al., 2022

McCarthy, A., Falloon, T.J., Danyushevsky, L.V., Sauermilch, I., Patriat, M., Jean, M.M., Maas, R., Woodhead, J.D., Yogodzinski, G.M. (2022) Implications of high-Mg# adakitic magmatism at Hunter Ridge for arc magmatism of the Fiji - Vanuatu region. Earth and Planetary Science Letters 590, 117592. https://doi.org/10.1016/j.epsl.2022.117592

), and suggests that in rare instances arc lavas retain a garnet-source signature without complete re-equilibration with the mantle wedge.

Full size image

In contrast to the typical island arc trend, increases in MnO/MgO values at continental arcs are much more limited (Fig. 4b), requiring significant garnet fractionation in addition to olivine ± clinopyroxene. It is particularly noteworthy that this divergent behaviour occurs during early fractionation of basaltic melts. Amphibole and/or Fe-Ti oxides also have low to very low SiO₂ contents and moderate to high K_D^Mn-Mg, and thus can also limit the extent of increases in MnO/MgO ratios (Figs. 4, S-2). However, these phases do not crystallise from typical arc melts until cooling to temperatures below at least ∼1050 °C and cannot be responsible for the divergent behaviour observed in basaltic compositions. Further, the minimum pressures at which garnet is stable increase for less evolved magmas (Fig. S-3). Early garnet fractionation from basaltic liquids therefore requires that garnet fractionation begins at pressures ≥1.5 GPa, or equivalently ≥50 km depth. Our current data do not allow us to distinguish between ubiquitous garnet fractionation and mixing between deep garnet fractionating magmas and more shallowly differentiating magmas. However, either scenario requires that at least some magmas undergo garnet fractionation at or below the Moho of many continental arcs.

The widespread fractionation of garnet-rich cumulates at or below the Moho in modern continental arcs is difficult to verify with geophysical techniques, as these cumulates have densities and seismic velocities comparable to or greater than sub-arc mantle (Müntener and Ulmer, 2006

Müntener, O., Ulmer, P. (2006) Experimentally derived high-pressure cumulates from hydrous arc magmas and consequences for the seismic velocity structure of lower arc crust. Geophysical Research Letters 33, L21308. https://doi.org/10.1029/2006gl027629

). Instead, fractionation of these cumulates near the base of the crust may contribute to the commonly poorly defined seismic Moho at many arcs. Finally, while both calc-alkaline (Fe-depletion) and tholeiitic (Fe-enrichment) differentiation sequences are observed at most arcs, calc-alkaline trends appear dominant in the thickest arcs (Chiaradia, 2014

Chiaradia, M. (2014) Copper enrichments in arc magmas controlled by overriding plate thickness. Nature Geoscience 7, 43–46. https://doi.org/10.1038/NGEO2028

). This observation, combined with the findings here, suggests an important role for garnet fractionation in the generation of some calc-alkaline differentiation sequences (e.g., Tang et al., 2018

Tang, M., Erdman, M., Eldridge, G., Lee, C.-T.A. (2018) The redox “filter” beneath magmatic orogens and the formation of continental crust. Science Advances 4, eaar4444. https://doi.org/10.1126/sciadv.aar4444

). However, calc-alkaline sequences are produced by extensive crystallisation differentiation that includes other phases such as FeTi oxides and amphibole. The stability of these phases is controlled by variables including fO₂ and magmatic H₂O contents, and thus a range of variables and fractionating assemblages likely combine to drive calc-alkaline differentiation (Sisson and Grove, 1993

Sisson, T.W., Grove, T.L. (1993) Experimental investigations of the role of H₂O in calc-alkaline differentiation and subduction zone magmatism. Contributions to Mineralogy and Petrology 113, 143–166. https://doi.org/10.1007/bf00283225

; Sisson et al., 2005

Sisson, T.W., Ratajeski, K., Hankins, W.B., Glazner, A.F. (2005) Voluminous granitic magmas from common basaltic sources. Contributions to Mineralogy and Petrology 148, 635–661. https://doi.org/10.1007/s00410-004-0632-9

; Zimmer et al., 2010

Zimmer, M.M., Plank, T., Hauri, E.H., Yogodzinski, G.M., Stelling, P., Larsen, J., Singer, B., Jicha, B., Mandeville, C., Nye, C.J. (2010) The role of water in generating the calc-alkaline trend: New volatile data for Aleutian magmas and a new Tholeiitic Index. Journal of Petrology 51, 2411–2444. https://doi.org/10.1093/petrology/egq062

).

top

Conclusions

On a global scale, trace element ratios such as Sr/Y and La/Yb appear to correlate with arc crustal thickness, and thus may be broadly reflective of varying extents of garnet fractionation. However, applying these proxies to individual magmatic suites can lead to incorrectly inferring garnet fractionation in magmas that instead reflect significant slab components and/or plagioclase accumulation. We show that MnO/MgO ratios avoid many of these limitations and provide a powerful tool to illustrate the role of garnet fractionation. Arc primary melts show a restricted range of MnO/MgO ratios that are consistent with melts in equilibrium with mantle olivine. Further, during early stages of fractionation, garnet is the only crystallising phase that does not drive magmas to evolve to higher MnO/MgO ratios. This diagnostic behaviour appears to be quite common in thicker continental arcs, suggesting that crystallisation differentiation with or without increasing fO₂ begins for many continental arc magmas at or below the Moho.

top

Acknowledgements

The authors thank P. Ulmer for providing us with experimental charges and M. Robyr and A. Ulyanov for assistance with EPMA and LA-ICP-MS respectively. Tom Sisson and Cin-Ty Lee provided valuable reviews of a previous version of this manuscript. OM acknowledges support from the Swiss National Science Foundation grant number 200021_197258 and BZK acknowledges funding support from the University of Lausanne.

Editor: Ambre Luguet

top

References

top

Supplementary Information

The Supplementary Information includes:

Analytical Methods

Tables S-1 to S-3

Figures S-1 to S-4

Data References for Figures 3 and S-2

Supplementary Information References

Download Table S-1 (xlsx)

Download Table S-2 (xlsx)

Download Table S-3 (xlsx)

Download the Supplementary Information (PDF)