BORON ISOTOPES ALONG AND ACROSS THE CENTRAL AMERICAN VOLCANIC ARC: NEW INSIGHTS FROM COSTA RICAN MELT INCLUSIONS

This work expands upon previous boron (B) isotope work on lavas from the El Salvador (Tonarini et al. 2007) and Nicaraguan segments (Turner et al. 2023) of the Central American volcanic arc, which demonstrate the utility of B to elucidate the rates and conditions of dehydration of the lower layers of the subducting Cocos plate. Here, we measured major and trace element abundances and B isotopic compositions (δ¹¹B) of olivine-hosted melt inclusions (MI) from an across-arc transect of Costa Rican tephras. Our new data expand spatial coverage across the arc and have the advantage of using naturally quenched MI. The resulting δ¹¹B measurements range from -8.42±1.32‰ to +2.56±0.94‰ with minor eruptive centers in the fore-arc and rear-arc, as well as within a few km of the arc-front, exhibiting lighter B isotope compositions relative to arc-front stratovolcanoes.

Trends in our Costa Rican arc δ¹¹B overlap with those from El Salvador and Nicaragua on a plot of δ¹¹B vs Ce/B, which means that the overserved variability of all three segments can be accounted for by a slab component with a δ¹¹B ranging from ~2‰ to ~6‰. As also found in Nicaragua, there are no correlations between δ¹¹B and sediment-controlled trace element ratios, indicating a deeper slab source for δ¹¹B from the ocean crust or slab lithospheric mantle. And as recently found in Chile (Turner et al., this session), δ¹¹B correlates better with distance from an arc-front stratovolcano than the distance across the arc.

The overlapping geochemical variations of the three arc segments are a surprising result, because the differing slab dip angles and ages might be expected to control the thermal structure of the descending slab, leading to differing B isotope fractionation and dehydration processes. To investigate why the δ¹¹B values of the slab component do not vary systematically, we developed new thermal and dehydration models of the subducting Cocos plate beneath each arc segment. The models make use of available constraints (and their uncertainties) on the initial compositions of slab lithologies on the down-going plate, the thermochemical structure of the subduction zones, and the expected fractionation of B isotopes as a function of temperature. Unlike previous work, our models use updated slab2 profiles and consider the uncertainties regarding the depth of coupling and the amount of shear heating along the slab-mantle interface.