The role of lower crust and continental upper mantle during formation of non-volcanic passive margins: evidence from the Alps OTHMAR MONTENER 1'2 & JORG HERMANN 1'3
llnstitut fiir Mineralogie und Petrographie, ETH Ziirich, CH 8092 Ziirich, Switzerland 2present address: Geology Institute, University of Neuch~tel, CH-2007 Neuch~tel, Switzerland (e-mail:
[email protected]) 3present address." Research School of Earth Sciences, ANU, Canberra, A.C.T. 0200, Australia Abstract: The remnants of a Mesozoic passive continental margin and of the Tethyan ocean floor are preserved in the Austroalpine and Upper Penninic nappes in eastern Switzerland and northern Italy. Reconstructions of the continent-ocean transition indicate that large areas of subcontinental mantle rocks, but only limited areas of lower-crustal rocks were exposed on the Tethyan sea floor. Microstrnctures, large shear zones, and the retrograde metamorphic evolution of peridotite and gabbro from Malenco (northern Italy) are investigated to evaluate the role of lower crust and upper mantle during formation of non-volcanic passive continental margins. The combination of petrological constraints and microstructures suggests two contrasting stages: (1) high-temperature (> 650°C) shearing and annealing of microstrnctures are attributed to pre-rift tectonics; (2) localized mylonitic shear zones cut the high-temperature structures and developed during nearly isothermal decompression (T 650°C) and later more localized deformation (T 650°C) in the lower crust and subcontinental upper mantle are probably older and not related to Jurassic rifting. A different scenario with a different P - T path for Early Jurassic rifting was published by Sanders et al. (1996) for mid-crustal rocks of the Southern Alps (Figs 1 and 9c) indicating a heating event at the onset of rifting. Bertotti et aL (1999) reached similar conclusions for rocks exposed along the Lugano-M. Grona line (Fig. 1). Through combined 4°Ar/39Ar, Rb/Sr and fission-track dating, Bertotti et al. (1999) argued that the thermal evolution of the South Alpine margin during rifting is mainly controlled by relaxation of isotherms that were assumed to be elevated because of inferred large mafic intrusions in the lower crust in mid- to late Triassic time (Bertotti & ter Voorde 1994)f Thermal gradients decreased from >60 °C kmat the initial stage of rifting to c. 20 °C km -1 after breakup. This decrease in the thermal gradient obtained by Bertotti et al. (1999) is opposite to what is preserved in Malenco and the Ivrea zone (Fig. 9a,b), where the thermal gradient increases because of near-isothermal decompression during rifting. Although there is field and isotopic evidence to support limited igneous activity during mid- and late Triassic time (e.g. Ferrara & Innocenti 1974; Stiihle et al. 1990; Lu et al. 1997), the modelled regional-scale heating in late Triassic time (Bertotti & ter Voorde
l
LOWER CRUST AND UPPER MANTLE DURING RIFTING 1994) is not supported by the work of Vavra et al. (1999) and Vavra & Schaltegger (1999). These workers argue that alkaline magmatism is limited and that widespread hydrothermal activity at 210 + 12 Ma caused alteration of zircons in Ivrea zone granulites. Such hydrothermal activity may well be associated with the onset of tiffing, and we propose that fluid accessibility is a major feature of early rifting in the deep crust. There may be spatial and temporal differences in how the deep crust and upper mantle respond to tiffing, but we favour the hypothesis that regional-scale heating was not important at the onset of rifting. We suggest that granulite-facies conditions in the lower crust are related to Early Permian decompression and associated magmatism, which was then followed by a period of near-isobaric cooling at all crustal levels. The P - T paths and their relation to rifting are especially important for unravelling the structural record of denuded subcontinental mantle where no lower-crustal rocks are attached. Microstructural analysis of exposed peridotites at the present-day West Iberian margin (e.g. Boillot et al. 1995a; Beslier et al. 1996) and ancient Adria (Erro Tobbio peridotite, Vissers et al. 1991, 1995; Hoogerduijn Strating et al. 1993) led to the hypothesis that both high-temperature and low-temperature structures were associated with rifting. This led to high-temperature, low-pressure P - T paths for these rocks (Fig. 9d,e) because in both cases plagioclase peridotites are preserved. A likely alternative is that only the latest ('hydrous?') part of the structural evolution is associated with rifting and that the anhydrous high-temperature evolution (i.e. >750°C) predates rifting, as it is the case in Malenco and the Ivrea zone. This hypothesis would imply that plagioclase peridotites are not necessarily related to rifting and that their role has to be reassessed. It remains to be tested if the plagioclase peridotites in the Alps (e.g. Lanzo) and in the West Iberian margin indeed represent processes related to rifting, or, more likely in our opinion, that they record infiltration processes into the subcontinental mantle. The lack of a genetic link between oceanic magmatism and underlying mantle rocks has been demonstrated by Rampone et al. (1995) for the Ligurian ophiolites and by Mtintener & Hermann (1996) for the Malenco peridotites, and is also the case in the Platta ophiolites (Miintener et al., unpubl. data). Metagabbros and pelitic granulites that were exhumed together with subcontinental mantle may help us to better understand the signifi-
281
cance of granulite-facies rocks along passive continental margins. Granulites are often considered to represent lower crust and their presence (or absence) has been used as argument to favour different tiffing models (e.g. Lemoine et al. 1987; Stampfli & Marthaler 1990; Froitzheim & Manatschal 1996; Froitzheim et al. 1996). The P - T estimates for the Malenco granulites and the Ivrea zone at the onset of rifting correspond to amphibolite-facies conditions. These rocks are only granulites because of high-temperature metamorphism and mafic underplating c. 50Ma before rifting. However, in areas without mafic underplating, the lowermost continental crust may consist of amphibolite-facies rocks that are hard to distinguish from upper-crustal basement rocks. On the other hand, as pointed out by Harley (1989) and Rudnick & Fountain (1995), not all exhumed granulites represent the lowermost continental crust. For example, zircon fissiontrack ages of granulites from the continentocean transition of the Galicia margin indicate that these rocks were already at upper-crustal levels before tiffing (Ftigenschuh et al. 1998) and do not represent the Mesozoic pre-tift lower crust. Granulite-facies rocks from the Valpelline series and the Sesia zone in the Western Alps (Fig. 9f) show a P - T path with near-isothermal decompression at much higher temperature than the Malenco granulites. For this reason, Gardien et al. (1994) favoured a late Variscan formation of the Valpelline granulites as a result of post-collisional thinning, and consequently exhumation of these granulites to mid-crustal levels is not related to Jurassic passive tiffing. These examples demonstrate that the presence of granulite-facies rocks next to passive margin sediments is not proof of exhumation of the lowermost crust during rifting, even though the Sesia and Valpelline rocks are in an Alpine tectonic position similar to the Malenco granulites. On the other hand, the metamorphic conditions in the lower crust at the onset of rifting reach only amphibolitefacies conditions. It is thus possible that parts of the lower crust are gneisses and amphibolites, which are not a priori recognizable as lower crust once they are exhumed during passive rifting.
A simple model f o r the formation o f passive continental margins
There has been a rapid shift in thinking about the formation of passive continental margins from essentially one-phase models (pure shear:
282
O. MUNTENER & J. HERMANN
e.g. McKenzie 1978; Le Pichon & Sibuet 1981; simple shear: e.g. Wernicke 1985; Lister et al. 1986; first applied to the Alps by Lemoine et al. 1987) to more complex two-phase models, which may include two simple shear systems separated in time (Hermann & Mtintener 1996), simple shear systems with spatially different dip along Tethyan passive margins (Stampfli et al. 1991; Favre & Stampfli 1992), or early pure shear followed by later simple shear along lowangle detachment faults with a top-to-the ocean sense of movement (Handy & Zingg 1991; Froitzheim & Manatschal 1996; Handy 1996; Manatschal & Bernoulli 1999). An important point is that kinematic inversions of seismic sections of the present-day Galicia margin (Manatschal et al. 2001) and reconstructions from the ancient Tethyan margins (e.g. Manatschal & Nievergelt 1997) show that the crust within the distal margin was already considerably thinned before detachment faulting became active. This indicates that detachment faulting is a late event in the formation of a passive continental margin, which can explain only a part of the thinning of the continental crust. Thus, most of the current debate considers the early rifting phase and the question of how lower crust and upper mantle are exhumed to shallow crustal levels, so that they can be denuded during late detachment faulting. The problem in the Alps is that the lower crust is exposed only in small parts of the Adria margin and not at all in the opposing European margin. Therefore most rifting models to date employ the polarity of facies distributions and sediment thickness in asymmetrical rifts to infer the dip directions of detachment faults at depth. For example, Schmid et al. (1997) invoked an earlier westand a later east-dipping detachment fault to explain the lack of lower crust in the Brianqonnals (European) margin. Yet, the continentward dipping Margna and Pogallo shear zones are the only geological evidence for large extensional shear zones in the lower crust along the ancient Tethyan passive margins. On the basis of the available geological evidence we propose that the scarcity of lower crust can be explained by the following simple model (Fig. 10). During an early rifting phase, stretching is accommodated by non-uniform large shear zones, which are probably symmetrically distributed on the scale of the whole rifted area, but are asymmetrical on the 1 0 - 1 0 0 k m scale (Fig. 10a). Seismic and field evidence (e.g. Bertotti et al. 1993) from sedimentary basins indicates that shear zones in the upper crust have a listric geometry and flatten at mid-crustal levels (at c. 10-
15 km depth), but are symmetrically distributed over the whole rifted area. On the other hand, shear zones in the deep crust and upper mantle uplifted lower-crustal and subcontinental mantle rocks to a depth of c. 10-15km. Although such a model could reconcile observations from the top 5 km of the crust with those from the lower crust, several important consequences must be addressed. (1) The site of major shear zones in the lower crust and upper mantle is laterally displaced from major tiffing at the surface. In this case, there must be a 'weak horizon' in the middle crust, where the theology of the crust changes significantly. This indicates that the upper crust may be decoupled from the lower crust and upper mantle by a weak mid-crustal layer. Such a weak layer might correspond to quartz-rich rocks, presumably at temperatures higher than 300°C (Handy 1987), around the brittle to plastic transition of quartz deformation. (2) The lower crust is considerably stronger than commonly assumed in many modelling studies that used quartz flow laws as an analogue for the lower crust (e.g. Dunbar & Sawyer 1989). In our model (Fig. 10a) the lower crust is 'boudinaged' if it is assumed that such large shear zones occurred on both future margins and a 'decoupling horizon' was present in the middle crust. During an advanced stage of rifting, extension was dominated by simple shear along lowangle detachments faults with a top-to-theocean sense of movement as documented and discussed by Froitzheim & Manatschal (1996) and Manatschal & Nievergelt (1997). The scarcity of lower crust along non-volcanic passive margins can be explained in three different ways: (1) the lower crust was exhumed close to the surface but remains covered by later emplaced blocks of upper continental crust (Manatschal & Bernoulli 1999); (2) the lower crust was not exhumed because detachment faults generally occurred in places where the lower crust was previously excised by localized shear zones of early rifting (lithospheric boudinage; see traces of future detachments in Fig. 10b); (3) the lower crust consists of amphibolites and gneisses and is therefore not easily distinguished from the mid- or upper crust unless detailed cooling ages combined with microstructures and thermobarometry are available. Although this is highly debated we favour the second hypothesis, because the geometry depicted in Figure 10 could explain the observation that subcontinental mantle is exhumed, but exhumation of the pre-rift lower crust on the ocean floor is extremely scarce and is so far demonstrated for only the Malenco area.
LOWER CRUST AND UPPER MANTLE DURING RIFTING
pre-rift
283
situation 0
30
~''_"
,-i
--,-'' (a),_.,, t- ( a } ' , ' - ~ ' - "
\~.'-
--~
