Oceanic lithosphere carries volatiles, notably water, into the mantle via subduction at convergent plate boundaries. This subducted water exercises a key control on the production of magma, earthquakes, formation of continental crust and mineral resources. However, identifying different potential fluid sources (sediments, crust and mantle lithosphere) and tracing fluids from their release to observed surface expressions has proved challenging 1 . The two Atlantic subduction zones are valuable end members to study this deep water cycle because hydration in Atlantic lithosphere, produced by slow spreading, is expected to be highly non-uniform 2 . As part of an integrated, multi-disciplinary project in the Lesser Antilles 3 , we studied boron trace element and isotopic fingerprints of melt inclusions. These reveal that serpentine, i.e. hydrated mantle rather than crust or sediments, is a dominant supply of subducted water to the central arc. This serpentine is most likely to reside in a set of major fracture zones subducted beneath the central arc over the past ~10 Myr. Dehydration of these fracture zones is consistent with the locations of the highest rates of earthquakes and prominent low shear velocities, as well as time-integrated signals of higher volcanic productivity and thicker arc crust. These combined geochemical and geophysical data provide the clearest indication to date that the structure and hydration of the downgoing plate are directly connected to the evolution of the arc and its associated hazards.The 750 km-long Lesser Antilles volcanic arc (LAA), located along the eastern margin of the Caribbean Plate, is the result of slow (1-2 cm/year) westward subduction of Atlantic and proto-Caribbean oceanic lithosphere (Fig 1). Water hosted in hydrous phases within the subducting plate will be released as the slab sinks into the mantle and warms up. As the water migrates out of the slab the stress on faults is reduced, causing earthquakes. At the same time, the addition of water to the overlying mantle wedge reduces the solidus temperature which may enhance melting. LAA magma production rates lie at the lower end of the global range, probably due to the low convergence rates, and are very unevenly distributed, being greatest in the centre of the arc (Dominica and Guadeloupe) 4 . The LAA also displays notable along-arc variations in geochemistry, volcanic activity, crustal structure, and seismicity [5][6][7][8] . Subducting plate velocity and age are often held responsible for variations in convergent margin behaviour 9 but are unlikely to have first-order influence on lateral variations within the LAA as neither vary significantly along-strike. Instead, variations in LAA magmatism and seismicity have been proposed to reflect; (i) a combination of a strong north to south increase in sediment input 10 , (ii) subduction of bathymetric ridges below the central arc 11 , which may enhance plate stress and coupling, (iii) and/or subduction of strongly hydrated fracture zones 12 at several locations along arc (Fig. ...

No abstract

No abstract

No abstract

No abstract

Volcanic rocks commonly display complex textures acquired both in the magma reservoir and during ascent to the surface. While variations in mineral compositions, sizes and number densities are routinely analysed to reconstruct pre-eruptive magmatic histories, crystal shapes are often assumed to be constant, despite experimental evidence for the sensitivity of crystal habit to magmatic conditions. Here, we develop a new program (ShapeCalc) to calculate 3D shapes from 2D crystal intersection data and apply it to study variations of crystal shape with size for plagioclase microlites (l < 100 µm) in intermediate volcanic rocks. The smallest crystals tend to exhibit prismatic 3D shapes, whereas larger crystals (l > 5–10 µm) show progressively more tabular habits. Crystal growth modelling and experimental constraints indicate that this trend reflects shape evolution during plagioclase growth, with initial growth as prismatic rods and subsequent preferential overgrowth of the intermediate dimension to form tabular shapes. Because overgrowth of very small crystals can strongly affect the external morphology, plagioclase microlite shapes are dependent on the available growth volume per crystal, which decreases during decompression-driven crystallisation as crystal number density increases. Our proposed growth model suggests that the range of crystal shapes developed in a magma is controlled by the temporal evolution of undercooling and total crystal numbers, i.e., distinct cooling/decompression paths. For example, in cases of slow to moderate magma ascent rates and quasi-continuous nucleation, early-formed crystals grow larger and develop tabular shapes, whereas late-stage nucleation produces smaller, prismatic crystals. In contrast, rapid magma ascent may suppress nucleation entirely or, if stalled at shallow depth, may produce a single nucleation burst associated with tabular crystal shapes. Such variation in crystal shapes have diagnostic value and are also an important factor to consider when constructing CSDs and models involving magma rheology.

No abstract

No abstract