The evolution of the Galápagos mantle plume

Abstract

The lavas associated with mantle plumes may sample domains throughout Earth's mantle and probe its dynamics. However, plume studies are often only able to take snapshots in time, usually of the most recent plume activity, leaving the chemical and geodynamic evolution of major convective upwellings in Earth's mantle poorly constrained. Here, we report the geodynamically key information of how the lithology and density of a plume change from plume head phase to tail. We use iron stable isotopes and thermodynamic modeling to show that the Galápagos plume has contained small, nearly constant, amounts of dense recycled crust over its 90-million-year history. Despite a temporal evolution in the amount of recycled crust-derived melt in Galápagos-related lavas, we show that this can be explained by plume cooling alone, without associated changes in the plume's mantle source; results are also consistent with a plume rooted in a lower mantle low-velocity zone also sampling primordial components.

Fig. 1.. Primary δ⁵⁷Fe throughout Galápagos plume evolution.

The correction to δ⁵⁷Fe_primary from the measured δ⁵⁷Fe depends on how the primary liquid MgO is calculated (Supplementary Text). The uncertainty on each point reflects both the range of primary MgO estimates, and the resulting uncertainty on δ⁵⁷Fe_primary (which depends on the slope of the δ⁵⁷Fe─MgO fit), with the typical analytical error shown separately. GSC, Galápagos Spreading Center; raw data from (18). See Supplementary Text for sources of bulk silicate Earth (BSE) isotope composition. Primary MORB liquids from (24, 39). Published mantle potential temperature (T_p) estimates are mostly calculated from major element chemistry (12, 13, 30, 33), and we use the middle panel to show the range of ages proposed for each locality (data sources for both age and temperature are given in table S2). We note that a consideration of harzburgite in the mantle source allows lower T_p estimates (31), but these estimates do not exist for all localities studied here. The bottom panel shows the calculated evolution of the pyroxenite fraction (5th and 95th percentiles of modelled solutions), with the minimum misfit solution in bold (see also Fig. 2). For the top and bottom panels, the data are plotted at the average age of the range shown in the middle panel.

Fig. 2.. Minimum misfit δ⁵⁷Fe_primary of the aggregate multilithology melt and pyroxenite fraction in the source from the Monte Carlo simulation.

The bold bars show the range of best-fit results for the range of bulk pyroxenite isotope compositions considered (δ⁵⁷Fe in purple, top; pyroxenite fraction in green, bottom). The paler envelopes show the 5th and 95th percentiles of the pyroxenite fractions (green, bottom) and resulting δ⁵⁷Fe_primary (purple, top) from the Monte Carlo runs accepted at 95% confidence, compared to the data and the peridotite-only (blue) case. The bottom panel also shows, in purple, the fraction of Fe in the aggregate melt derived from pyroxenite. The pyroxenite fraction results are repeated from Fig. 1 to highlight the balance and distinction between changes in fraction of pyroxenite, and fraction of Fe derived from pyroxenite, as the plume evolves.

Fig. 3.. Constraints from pyroxenite entrainment fraction on the density of entrained material.

(A) Modified from (14), estimates of maximum entrainment fraction of dense material in the plume head for different boundary layer thicknesses of that material. The grey bar shows the buoyancy number implied by the range of minimum misfit pyroxenite fractions in Tortugal for the range of bulk pyroxenite isotope compositions considered (purple shading). Buoyancy number, B, is defined as ∆ρ/(α ∆T ρ_c); Supplementary Text. (B) Estimate of excess density (red lines) of entrained material in the Galápagos plume relative to ambient mantle, based on B = 0.8 and using suitable estimates of thermal expansion, α and driving temperature contrast, ∆T = T_CMB − T_p (see section S7). The contours mark density excesses in α-∆T space, as shown by the color bar. The solid red line shows density results using the ∆T estimate using the Tortugal T_p, and the red dashed line shows results using the ∆T estimate considering non-adiabatic cooling on the lower mantle T_p of Tortugal (section S7).