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

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.

The six factors also agree with the mixing trends observed in element x vs. y plots and ternary diagrams (Supplementary Fig. 1 and 2).
If we force an iteration of the QFA to explain the dataset variability with seven factors, the seventh factor has high factor scores for Al, Ti, Rb, and Cs, and expresses a factor score pattern similar to the mafic basalt end-member defined in the aluminosilicate QFA of Dunlea et al. (2015a) 1 . However, because it explains only < 1% of the variability, we consider it statistically insignificant and ignore it in this study.
accounting for 0.6 to 1.5 Tmol/yr of the total Mg uptake 8 . As is reported in the main text, our calculations suggest that only 0.02 Tmol/yr are taken up into global pelagic sediment.
Assuming steady state, this is based on parameters typical of the SPG pelagic clays, including a dry bulk density of 0.35 g of bulk sediment per cubic centimeter 39 , a bulk sediment that is comprised of ~20% of an ash enriched in Mg by 3.9 wt.% beyond the expected value for the precursor unaltered ash 1 , an average sedimentation rate of 1 m/Myr 10 , and the global spatial area of authigenic clay-bearing pelagic sediment of 1.8 x 10 8 km 2 (~50% of the total seafloor) 11 . Thus, deep-sea pelagic sediment, the ocean's most spatially extensive sedimentary lithology, currently is not a significant sink of Mg in the modern ocean.
For the calculation of Mg uptake into more Si-rich sediment, we considered the characteristics of the Si-enriched bulk sediment sample modeled in this study closest to the chert layer at Site U1365 (at 73.9 mbsf). The high fraction of Mg-enriched altered ash and relatively fast accumulation rates indicate more Mg being taken up into clays.
Assuming this sample approximates Si-rich deposits from the early Cenozoic (e.g., 12,13) , we changed the parameters in the previous calculation to be similar to this Si-enriched sample. The calculation used a dry bulk density of 0.4 g of bulk sediment per cubic centimeter 9 , 66% of the sediment being Mg-enriched altered ash, and a sedimentation rate of 5 m/Myr. With this type of sedimentation occurring across 50% to 100% of the seafloor, 0.4 to 0.8 Tmol of Mg would be removed from seawater every year. This is 25-33% of the modern riverine influx of Mg from silicate weathering and is the same order of magnitude (~Tmol) as that required to drive the observed increase in seawater Mg/Ca over the Cenozoic 8 .

Supplementary Note 3. Calculating changes in atmospheric CO 2 caused by changes in reverse weathering
Our results suggest that the Early Eocene had an additional 6 x10 12 alkalinity equivalents/yr (alk eq/yr) from silicate weathering. Exactly how much higher atmospheric CO 2 and temperatures needed to be in the Early Eocene to increase chemical weathering by that amount depends on a large number of factors. Here, we perform a first-order estimate by assuming a global carbon cycle at steady state and silicate weathering that depends on pCO 2 according to: (2) where and are the silicate weathering flux and atmospheric CO 2 in the modern era, 24 x 10 12 alk eq/yr and 280 ppm, respectively. is a constant that represents the complex set of processes that link atmospheric CO 2 and global silicate weathering rates (i.e., temperature, runoff, 'weatherability', etc) [14][15][16][17] and is set at 0.3, consistent with previous studies 18 . We estimate , the rate of silicate weathering in the past, to be 30 x 10 12 alk eq./yr. That is, 6 x 10 12 alk eq/yr higher than the modern era to balance the higher rates of reverse weathering. Solving for shows the change in reverse weathering results in an approximate doubling of atmospheric CO 2 from 280 ppm as seen in the modern era to 589 ppm. In the above calculations and plots, we are assuming that the Mg in the original volcanic ash shard is retained in the authigenic phase. If the volcanic glass shard is completely recrystallized on the seafloor, 100% of the Mg in the authigenic mineral will be from seawater and have a δ 26 Mg of a secondary mineral forming from seawater. With this conceptual model and the other evidence that ~100% of the Mg is from seawater, we can use the δ 26 Mg of our Mg-ash layer analysis (0.52 ‰) and state that it is fractionated +1.34 ‰ from seawater composition (-0.82 ‰). This is very close to, but slightly higher than, the fractionation predicted from porewaters (+1.25 ‰) 24 .

Supplementary Figure 5. Mg concentrations and δ 26 Mg of interstitial water.
Concentrations of Mg (mM) 9