The geochemical pattern of the hydrous eclogites from Trescolmen permits their use as a case study for the behaviour of low-T altered basalts undergoing subduction. We have shown evidence for fluid migration under high pressure conditions (>1.8 GPa). The total amount of fluid could have been very low. Nevertheless one scenario can be excluded: If fluid from the surrounding metapelites was just enough to form amphibole, the resulting Cs/Rb ratio (ca. 0.2) would be almost two orders of magnitude higher than observed (0.004 for sample 59-1). The most efficient mechanism to explain the observed LILE pattern of the eclogitic amphiboles is pervasive fluid circulation equivalent to open system behaviour. The fluid in equilibrium with the surrounding metapelites infiltrates the eclogite bodies, in case of 59-1 at least on a metre scale. The calculated W/R ratio is extremely low (from 0.01 to 0.1).
W/R ratios of as low as 0.01 are in accordance with observed oxygen isotope heterogeneities on a decimetre scale in eclogites from the Tauern Window in a structurally similar setting (Getty & Selverstone, 1994), since oxygen isotope changes would be too small to be measurable. Fluid-rock ratios of 0.01 will cause only minor changes in δ18O values even for interaction between two extremely different compositions (e.g. fluids in equilibrium with carbonate sediments interacting with mantle-derived basalts would shift δ18O by 0.2 ‰). Fluid interaction between less extreme compositions such as oceanic pelites and low-T altered basalts (Trescolmen case) would result in almost undetectable changes in the δ18O values, even for W/R ratios of 0.1. Experimental results of Watson &
Lupulescu (1993) indicate non-wetting behaviour of aqueous fluids in clinopyroxene-rich systems which seems to be in contrast with our findings. Nevertheless, the strong anisotropy induced in rocks undergoing plastic deformation may facilitate fluid influx, as discussed for mylonitised marbles by Holness (1997). This effect is likely to dominate in most eclogites from Trescolmen, where a penetrative foliation is locally defined by partly aligned amphiboles.
Since fluids are able to infiltrate eclogitic rocks under high pressure, it is concluded that fluids can also leave this system. The ability of very small volumes of fluid to penetrate eclogitic systems is a prerequisite of many models addressing the large-scale behaviour of subduction zones, although the migration mechanisms and potential problems with fluid migration are seldom considered. The existence of volcanoes above subducting slabs of variable depths (from ca. 120 km to >400 km), together with systematic geochemical changes in these volcanic rocks, has been explained by a steady, though decreasing amount of fluid flux from the subducting slab wedge into the underlying mantle (Woodhead & Johnson, 1993; Ishikawa & Nakamura, 1994; Ryan et al., 1995; Ishikawa & Tera, 1997; Shibata & Nakamura, 1997; Moriguti & Nakamura, 1998). Most of this fluid component (as high as 99% in the case of the Izu arc) is thought to originate from the upper part of the altered oceanic crust, based on Li-B-Pb isotope systematics (Moriguti & Nakamura, 1998). The amount of fluid available from continuous and discontinuous dehydration reactions has been modelled by Poli & Schmidt (1995) and Schmidt & Poli (1998). In cold subduction zones, only about 0.1 wt% of H2O is produced per 50 km depth, mostly from the continuous reactions lawsonite + diopside + garnet1 -> garnet2 + coesite/stishovite + H2O, and phengite -> K-rich clinopyroxene + enstatite + coesite + H2O (Schmidt & Poli, 1998). These small amounts of fluids must be able to leave their eclogitic source rocks in a more or less continuous manner to ultimately produce the observed across-arc variations in many arc systems (Kuriles, Izu-Bonin, New Britain, NE Japan, etc.). Our results support such a scenario in which small amounts of H2O can leave their production site, probably facilitated by ductile deformation, and subsequently becoming collected in an anastomosing vein system (Widmer, 1996).
2.6 Conclusions
1. Eclogites from Trescolmen show petrographic evidence for fluid migration under high pressure conditions.
Kyanite is partly replaced by paragonite, barroisitic amphibole overgrows older omphacite and high pressure (>2.0 GPa) talc veins infiltrate a Mg-rich eclogite body.
2. LAM-ICP-MS measurements on all high pressure mineral phases reveal that phengite is the dominant phase for Cs, Rb and Ba in eclogitic rocks, accommodating >90% of the whole rock budget in phengite-bearing samples.
Other phases such as paragonite and amphibole have only a very limited capacity to incorporate these elements.
3. Comparison of whole rock and in-situ mineral data shows that secondary phases, even in the freshest rock samples, can severely alter the composition of whole rock samples for fluid-mobile elements such as Ba. This clearly demonstrates the advantage of in-situ geochemical methods over whole rock analysis in extracting petrogenetic information in complex metamorphic assemblages.
4. Cs/Rb and Ba/Rb ratios of eclogites and garnet micaschists retain the patterns of their likely protoliths, i.e. low-T altered basalt and upper continental crust, respectively. LILE pattern of amphibole in phengite-lacking samples plot outside any likely protolith, but are identical to patterns in amphiboles in phengite-bearing samples. The observed data can be explained by flow of fluid derived from metapelite through the eclogite bodies, leading to amphibole homogenisation. Nevertheless the amount was small enough to leave original phengite patterns intact.
5. Modelling of open system fluid-rock interaction give a range of possible W/R ratios of 0.01-0.1. This result is consistent with the observation of oxygen isotope heterogeneities in other alpine eclogite bodies, showing that in-situ LILE analysis are more suitable for regions of low W/R interaction.