Espectroscopía ultravioleta-visible (UV-VIS)

In most mafic and metasedimentary samples, garnet is not zoned in δ18_{O. However, four mafic}

samples show δ18_{O zoning linked to the presence of Mg-rich rims. Garnets yield generally high}

δ18_{O values up to 14.7 ‰. The lowest value is found in eclogite SHB12B (6.2 ‰), other}

metabasites yield high values that converge toward the value measured in the quartzite SHS03 (14.6 ‰).

Mg-poor outer garnet rims are observed in 5 mafic samples (SHB12B, SHB45, SV01-75, SV12- 13F and SV03-103, Figure 4 - 23). These rims grow from a dissolution surface that is irregular,

and follows cracks and inclusions into the garnet. In some samples, this rim is large (up to 300 μm width in SHB45, replacing >50% of the garnet), but in others it is partial and small (20 μm) e.g. SV01-75. The size of the rims is different in different compositional sections in SV03-103: garnet in the omphacite-rich zones have 100 μm rims whereas the Mg-poor garnet rim is almost absent in the lawsonite-rich section. In places, the rim growth is coupled to dissolution of the pre- existing garnet as is shown by Rayleigh-like depletion in Mn, Y and HREE at the core-rim interface (e.g. profile in the core crack of SHB12B garnet, Appendix figure A4 – 8) in the inner part of the rim. This re-incorporation of garnet-compatible elements is interpreted as a loss of garnet volume due to dissolution. In other grains and samples, it is not clear if the rim is new growth around dissolved garnet or is replacing garnet core, as the profiles are flat for garnet- compatible elements at the interface. The rim shows euhedral edges indicating growth, at least in samples where garnets is not affected by chloritization (e.g. SV12-13F), Most rims show an enrichment in MREE and sometimes HREE, which could either be brought by the fluid, or reflect the dissolution of REE-bearing phases such as garnet, allanite and lawsonite, equilibrated at the scale of the sample.

Figure 4 - 23. Variation of δ18_{O according to garnet Xgrs and XMg. Ties indicate zoning as observed} in single grains.

These garnet rims usually show a δ18_{O that is heavier than the garnet cores it surrounds,around}

11-12 ‰ in samples SV12-13F, SHB12B and SV01-75 with the exception of outer rim in (up to 13 ‰) eclogite SHB12B, ca. 14 ‰ in sample SHB45 and up to 14 ‰ for the few small rims in SV01-75. The high δ18_{O garnet rim in all samples has a rather homogenous Mg# (10-15, Figure}

4 - 23), in contrast to the cores which show more variability in composition across samples (8- 25). If one takes Mg# as a relative indication for temperature, the homogeneous composition suggests that the rims crystallised at similar temperatures in all samples, with Mg# higher than core in SV01-75 and SV03-103, and lower than core in SHB12B and SHB45. The core-rim contrast seen in SHB12B at ca +6‰ is among the largest δ18_{O contrasts reported for garnet}

(Vielzeuf et al. 2005a; Page et al. 2014; Martin et al. 2014b). The texture of the rims is similar to the garnets in lawsonite schist 09028 of Martin et al. (2014b) and the garnets from a Franciscan amphibolite (Page et al. 2014b) but have a reverse δ18_{O behaviour, indicative of a different fluid}

source than in other studies, with a more prominent sedimentary influence (see below).

Additionally to the Mg-poor rims, garnet textures such as re-equilibration channels transitioning to Mg-rich rims are observed in samples SV03-103 and SV12-13F, as well as a Mg-rich mantle in SV01-75. They resemble what has been described in eclogite garnets of several localities as well as amphibolites (e.g. Whitney et al. 1996), sometimes linked to atoll textures (e.g. Cheng et al. 2007). In these instances, sub-grain boundaries or fluid channels allow partial replacement of the major elements in garnets typically with a change in Mg# (Konrad-Schmolke et al. 2007; Faryad et al. 2010). These textures are interpreted as a re-equilibration mechanism of the prograde cores when increasing temperature at thermal peak or the start of exhumation, with the catalysing presence of fluids. In SV12-13F, a δ18_{O contrast is present between the core that yields a high}

δ18_{O at ca. 14 ‰ and the cross cutting zones at ca. 11 ‰. In the preliminary data obtained for}

SV03-103, no such contrasts can be observed since the cores also yield 10 ‰ on average (Appendix figure A4 – 15). Both of these mantles thus seem to have equilibrated with a fluid of ca. 11-12 ‰, although the fluid can only be identified in SV12-13F, as the core of the garnet records a contrasted bulk composition.

In SV03-103, a lawsonite-rich layer interpreted as a vein by Davis and Whitney (2008) is present, in which garnet and lawsonite cores display oscillatory Ca, REE and Cr zoning, which have elsewhere been interpreted as the result of fluid-influx-related disequilibrium (Sherlock 1999). SV03-103 vein garnets also contains inclusions of rock-forming minerals that are not present in the matrix (quartz, followed by omphacite towards the rim). Such inclusions are likely to be the result of prograde growth and are not observed in other samples of this study. These oscillatory zones could thus be the result of prograde fluid circulations in this rock. These two samples are taken from the pod described in Davis and Whitney (2008): SV03-103 is sampled from the edge of the pod, and SV12-13F from the core. In SV03-103, garnets in the vein as well as the matrix yield a homogenous garnet δ18_{O of ca. 9-10 ‰, similar to the signature observed in the garnet}

in vein minerals is an indicator of protracted fluid influxes, it is possible that the SV03-103 area of the pod was equilibrated with lower δ18_{O fluids prior to the start of the garnet crystallisation,}

whereas the core of the pod retained high values until a more pervasive fluid event at peak conditions.

In summary, garnet in several mafic samples show the influx of high- δ18_{O fluids at two stages:}

during prograde-peak metamorphism (Mg-rich re-equilibration channels) and at the start of exhumation (dissolution of prograde garnet and growth of a Mg-poor rim). This change in δ18_O

is not followed by systematic changes in garnet-compatible trace-element concentrations, which are likely linked to local redistribution.