Field observations and electron micro-probe analyses indicate that pseudotachylytes discovered on the Lofoten island of Flakstadøy, north Norway, represent rare examples of deep-crustal paleoseismic faults. The pseudotachylyte occurrences are restricted to the margins of eclogite-facies shear zones that sharply cut pristine granulite-facies continental basement rocks. Generally, pseudotachylyte veins are sharply truncated by the eclogite shears, but some have been sheared and folded into them, documenting prekinematic to synkinematic injection. Textures preserved in the pseudotachylyte matrix document crystallization directly from the frictional melt; for example, dendritic garnets, similar in appearance, size, and composition to those from eclogite pseudotachylytes of the Bergen Arcs and Ålesund (Austrheim and Boundy, 1994; Lund and Austrheim, 2003), reflect rapid (likely in terms of tens of seconds) crystallization, and distinct fining of grains toward the margins of the pseudotachylyte veins indicates quenching textures. Electron microprobe analysis and backscattered-electron imaging document that the pseudotachylyte matrix is composed of microlites of garnet (Gr25–30, Py15–19, and Al54–58), orthopyroxene (En61–64), low-Na clinopyroxene (Jd6), amphibole (ferroan pargasite), with or without K-feldspar, quartz, biotite, various Fe opaques and Fe-Ti opaques, kyanite, dolomite, and calcite. The cogenetic eclogite-facies shear zones and pseudotachylytes were variably retrograded during Caledonian amphibolite-facies metamorphism. Omphacite is replaced by clusters or symplectites of low-Na clinopyroxene (Jd6) and oligoclase/andesine (An20–36); kyanite, orthopyroxene, Na-Ca clinopyroxene, amphibole, and dolomite occur as inclusions in garnet. The Flakstadøy pseudotachylytes indicate that the rocks exposed in Lofoten were rigid and resilient parts of the lower crust of an ancient continent from ca. 1.8 Ga until the Middle Ordovician. Subduction to deeper-crustal levels (depths >∼45 km) caused the stiff, nonreacted granulite to accommodate aseismic, steady-state flow in fluid-mediated, eclogite shear zones by concomitant, brittle, seismogenic failure and pseudotachylyte formation. Later in the Middle Ordovician, these deep-crustal rocks were exhumed to middle-crustal levels, where they were retrograded under amphibolite-facies conditions. Our results help to explain how deep-crustal earthquakes form in modern continent-continent collisional zones like the Himalayas.

One of the more visual and elucidating tenets of plate-tectonic theory is the lateral and vertical progression of earthquake foci and their correlation to depth along subduction zones (Wilson, 1963). Deep-foci earthquakes (between 300 and 680 km deep) require deep subduction of oceanic lithosphere (e.g., the Philippines; Hurukawa and Imoto, 1993). Since the ultimate fate of the deeply subducted oceanic lithosphere is its consumption and recycling within the mantle, no surface exposures containing evidence of deep-focus paleoseismic faulting of this material are known. Hence, our understanding of such phenomena is fragmentary and based on inferences from indirect geophysical, mostly seismic, information. On the other hand, subducted continental crust, being less dense and consequently more buoyant, is less likely to be wholly consumed, affording the potential for exhumation and exposure of the actual rock products of deep-crustal seismic faulting (e.g., the Himalayas; Kayal et al., 1993; Jackson et al., 2004). Thus, eclogitized continental basement exposed in the cores of ancient eroded mountain belts are fertile grounds for exploring such rock products. Vast exposures of high- and ultrahigh-pressure eclogitized continental basement of the Western Gneiss Region of western Norway make it a prime area in which to explore for deep-focus paleoseismic faults. The island of Holsenøy in the Bergen Arcs (Austrheim and Boundy, 1994) and Ålesund (Lund and Austrheim, 2003) contain, to the best of our knowledge, Earth's only known examples of high-pressure pseudotachylytes (Fig. 1).

We report the discovery of pseudotachylyte veins associated with eclogite-facies shear zones within continental basement in the Lofoten archipelago, north Norway (Fig. 1), which also appear to be deep-crustal paleoseismic faults. Pseudotachylytes are largely accepted as frictional melts derived from coseismic faulting and are the only known recorders of the process preserved in exhumed rocks (Shand, 1916; Philpotts, 1964). The particular group of pseudotachylytes described herein is restricted only to the immediate shoulders of eclogite-facies shear zones, a field relation that in itself seems to require a cogenetic, deep-crustal origin (Steltenpohl et al., 2003; Kassos et al., 2003, 2004). We present petrographic observations, electron microprobe analyses, and backscattered-electron images (BSEs) documenting that original eclogite-facies assemblages and textures were quenched within the pseudotachylyte matrix prior to being strongly retrograded under amphibolite-facies conditions. Available timing constraints suggest that the eclogite-facies pseudotachylytes and shear zones, and their subsequent amphibolite-facies retrogression, occurred during the early stages of the Caledonian orogeny (Steltenpohl et al., 2003; Kassos et al., 2003, 2004). Our findings, thus, bear on how earthquakes, like those beneath the active Himalayas, are generated in the deep levels of the continental crust, where high temperatures and pressures prevail and plastic, aseismic failure might be expected.

A transect across the Caledonian orogen along latitude 68.5°N, Norway (A–A′ in Fig. 2), exposes subequal proportions of Baltic Pre-cambrian continental basement and its cover, and may represent a continuously exposed column through nearly the entire Caledonian lithosphere. This transect progressively traces the basement-cover contact, from east to west, from the unmetamorphosed and undeformed nonconformity in the Swedish foreland, to a greenschist-facies mylonite zone framing the external Rombak window, to an upper amphibolite–facies mylonite zone against the Western Gneiss terrane, across the amphibolite-granulite isograd (i.e., the Conrad discontinuity; Olesen et al., 1991), and finally into eclogite-facies rocks in the westernmost Lofoten Islands (Fig. 2). The eastern half of this transect is well characterized and accepted as the middle- to upper-crustal, Caledonian (Silurian), continental-continent subduction zone boundary, where western Baltica was partially subducted beneath Laurentia (Hodges et al., 1982; Tull et al., 1985). The Lofoten block, on the other hand, is a long-held enigmatic terrane with uncertain relations to both Baltica and the Caledonides (Griffin et al., 1978; Tull, 1977; Tull et al., 1985; Olesen et al., 1997; Klein and Steltenpohl, 1999).

Lofoten is composed of Archean (2.7 Ga) rocks migmatized at ca. 2.3 Ga, supracrustals deposited ca. 2.1 Ga, and extensive mangeritic and charnockitic plutons emplaced under granulite-facies conditions between 1.8 and 1.7 Ga (Griffin et al., 1978). Structurally isolated bodies of amphibolite-facies metasedimentary rocks, the Leknes Group (Fig. 3), are interpreted as early Caledonian (Ordovician-Silurian) klippen preserved in down-folded and faulted structures (Tull, 1977; Corfu, 2004; Steltenpohl et al., 2004). In Carboniferous plate reconstructions, east Greenland is welded to the Norwegian margin, and Lofoten clearly occupied the most internal tectonic position within the northern parts of this restored orogen (Fig. 1). Surprisingly, however, little Caledonian imprint is preserved in the Lofoten basement rocks. Caledonian structures and fabrics along the base of the cover allochthons gradually disappear structurally downward into the basement over a distance of ∼250 m (Tull, 1977), leaving earlier workers to suggest that Lofoten had “completely escaped Caledonian metamorphism and deformation” (Griffin et al., 1978). Two hypotheses have been suggested to explain these observations. First, the Caledonian allochthons passed over Lofoten, and the downward disappearance of Caledonian fabrics and structures may be attributed to the limited availability of fluids in the anhydrous, granulite-facies basement (Bartley, 1982; Steltenpohl et al., 2004). Second, Lofoten might be a beached microcontinent (Tull, 1977; Corfu, 2004). Most workers favor the first interpretation (see Hodges et al., 1982, and Steltenpohl et al., 2004). However, our present understanding of the timing and structural evolution of Lofoten and its contact with Baltic crust (that is, the Gullesfjorden shear zone and/or the Austerfjord thrust in Fig. 2) is fragmentary, leaving the problem unresolved (see Hakkinen, 1977; Tull, 1977; Corfu, 2004).

Rare eclogite-facies shear zones that sharply cut the granulite-facies gneisses occur on the islands of Austvagøy and Flakstadøy (Figs. 2 and 3) and appear to be expressions of early Caledonian deformation (Steltenpohl et al., 2003; Kassos et al., 2004; Rehnström et al., 2005). Kullerud (1992, 1996) and Markl and Bucher (1997) performed detailed petrologic and mineral chemical studies on the Lofoten eclogites, and reported that they are variably to completely replaced by amphibolite-facies assemblages. The same authors surmised that the eclogites had formed due to fluids having accessed fractures into the anhydrous, granulite-facies basement units. This unusual style of occurrence is remarkably similar to eclogites found in the Bergen Arcs (Fig. 1; Austrheim, 1987; Austrheim and Griffin, 1985; Boundy et al., 1992). Shear-zone eclogites in the Bergen Arcs and Lofoten thus are important evidence of fluid flow in the deep crust (Austrheim, 1987; Austrheim and Griffin, 1985; Boundy et al., 1992; Kullerud, 1996, 2000; Markl and Bucher, 1998; Markl et al., 1997, 1998a, 1998b; Bjørnerud et al., 2002). The Lofoten eclogites further resemble those of the Bergen Arcs in that: (1) they formed prior to retrograde amphibolite-facies metamorphism before ca. 433 Ma and, thus, appear to be early Caledonian (Mørk et al., 1988; Steltenpohl et al., 2003) rather than Scandian (425–400 Ma) eclogites like those of the classic Western Gneiss Region (Fig. 1; Griffin and Brueckner, 1980; Hacker et al., 2003; Terry and Robinson, 2003); (2) both cut Archean-Proterozoic orthogneisses and associated granulite-facies gabbroic and anorthositic rocks; and (3) pressures of eclogitization in these two areas are much less (Lofoten ∼1.4–1.5 gPa; Bergen Arcs ∼1.7 gPa) than the ≤4 gPa estimated for the minimum pressures of the Western Gneiss Region (see Austrheim, 1987; Markl and Bucher, 1997; Steltenpohl et al., 2003). Compared to the Bergen Arcs, however, the Lofoten eclogites were much more intensely retrograded during Caledonian amphibolite-facies metamorphism, leaving us with a relatively fragmented understanding of eclogitization and eclogite shear-zone development in Lofoten.

Despite their obvious significance for deformation in the deep-continental crust, to our knowledge, we and our co-workers are the first to perform structural investigations (i.e., geometric, kinematic, and microstructural) on the Lofoten eclogite shear zones (Kassos et al., 2003, 2004, 2005; Mager et al., 2004). It was during our field investigations of the shear zones that we stumbled upon the associated pseudotachylyte veins that are the focus of the present report. The structural evolution of the eclogite shear zones is complex and beyond the scope of the current report. To summarize pertinent observations, the eclogites are concentrated on the island of Flakstadøy (Fig. 3), where they occur in relatively small, localized areas that range from 40 m2 (Nusfjord) to 1.6 km2 (Skagen). Individual eclogite occurrences do not connect with one another, and there does not appear to be any particular tectonostratigraphic level or zone to which they belong. The shear zones may occur individually, ranging in thickness from a millimeter to <10 m, or as anastomosing networks up to 100 m in aggregate thickness. Strikes are highly variable, encompassing almost all directions, and dips range from vertical to subhorizontal. Most are simple shear zones with clear kinematic indicators (e.g., S-C fabrics and asymmetric porphyroclasts) that display highly variable movement directions, even within individual outcrops. Nonfoliated granulitic host rock commonly is progressively foliated toward the margins of the shear zones (Fig. 4), and this foliation commonly is asymptotically swept into them. Displacements along individual shear zones, based on displaced markers and the observation that both terminations of some shear zones occur within a single outcrop, typically are small, ranging from negligible to a few centimeters. An “eclogitization front” may extend for tens of centimeters outside the shear-zone margins (Fig. 4). The shear zones commonly branch and merge or crosscut one another (Fig. 4).

At the time this report was written, the timing of eclogite-facies shear-zone formation in Lofoten was only loosely constrained but likely resulted from early Caledonian (Middle Ordovician–Early Silurian) orogenesis. Corfu (2004) reported U-Pb mineral dates on zircon and titanite from meta-igneous rocks within the Leknes Group (Figs. 2 and 3) that are interpreted to bracket the time of the amphibolite-facies event between 461 and 469 Ma. We interpret this to be the same amphibolite-facies event that retrograded the Lofoten eclogites. This is consistent with a ca. 433 Ma 40Ar/39Ar cooling date on hornblende separated from a sample of the retrograded eclogite at Nusfjord (Steltenpohl et al., 2003). Kassos et al. (2004) reported a 478 ± 41 Ma lower-intercept age from U-Pb analysis of zircons separated from a pre-eclogite-facies felsic dike from the eclogite shear zone at the Myrland locality (Fig. 3). This date carries a large error but is compatible with eclogitization just before amphibolite-facies metamorphism of the Leknes Group at ca. 469 Ma.

Pseudotachylyte veins are associated with some eclogite-facies shear zones at the Nusfjord and Skagen localities on Flakstadøy (Fig. 3). The pseudotachylytes do not occur outside of the immediate area along the contacts of the eclogite shear zones. We only find pseudotachylytes where there are eclogite shear zones, but, conversely, most of the eclogite shear zones do not have associated pseudotachylytes. The pseudotachylytes occur as thin (≤3 cm thick), mostly tabular veins along the shoulders of <2-m-thick eclogite shear zones that cut gabbronorite of the basement complex (Figs. 4 and 5A–C). Gabbronorite host rock is composed of plagioclase (An50–65), orthopyroxene, clinopyroxene (sub-calcic augite), magnetite, ilmenite, and apatite, with grain size ranging from 0.5 to 1 cm (Kullerud, 1992, 1996). Although the mineralogy of the pseudotachylyte matrix, described in the following, contrasts with that of its granulite-facies host rock, the mineral chemistries and their estimated volume percentages are consistent with the two rock types being of essentially the same chemical composition. Combined with their aphanitic character, clear field association restricted to the margins of shear zones, and lack of field evidence to the contrary, the veins clearly are pseudotachylytes derived from frictional melting of the gabbronorite.

The pseudotachylytes are dense, dark greenish gray to black, microcrystalline rocks that generally occur as thicker (<3 cm) tabular veins with smaller, thinner (only millimeters thick) wedge-shaped veins branching off of them (Figs. 5A, 5B, and 6). Inclusions of gabbronorite host rock are common in the veins (Figs. 5B and 6). Where we were able to observe their interaction, most pseudotachylyte veins are abruptly truncated by the eclogite shears (Fig. 5A). Importantly, there is no corresponding “other half” of the vein on the opposite block, even where displacement along the shear zone is demonstrably negligible. This latter observation seemingly requires a cogenetic relation since the veins clearly sourced or fed from the shears and did not simply behave as passive markers that were crosscut by the shears. Many pseudotachylyte veins parallel the boundaries of the eclogite shear zones (Fig. 4). Thin (<10 cm thick), small-displacement (<10 cm) shear zones may have pseudotachylyte veins in their centers (Fig. 4), clearly indicating that they had nucleated along them. Rarely, pseudotachylyte veins have been sheared, folded, and dragged into the eclogite shears (Figs. 6 and 7). These sheared pseudotachylytes were only observed in the marginal areas of some thicker (3–4 m thick) eclogite shears. The veins progressively lose definition toward the more highly strained centers of the shear zone. Figure 6 illustrates a spectacular example where pseudotachylyte veinlets branching off of a thicker vein, which parallels the shear zone (in the C-plane orientation), have been only slightly sheared into parallelism with the S-plane orientation of the shear-zone system. Taken together, field relations clearly indicate that the pseudotachylytes and the eclogite shear zones formed cogenetically, and that the former slightly predated or temporally overlapped with development of the latter.

Microlites within the pseudotachylyte matrix are generally too fine grained to confidently resolve using an ordinary petrographic microscope. Microprobe analyses (01), therefore, were performed using facilities at the University of Alabama. We probed three different thin sections from three separate pseudo tachylyte veins at the Nusfjord locality (NFA-11, NFA-18, and NFA-20). All three samples show high degrees of amphibolite-facies retrogression.

The matrix of the pseudotachylyte is an ultrafine-grained mosaic of (in decreasing volumetric abundance based on visual estimations) plagioclase, amphibole, garnet, orthopyroxene, clinopyroxene, Fe oxides and Fe-Ti oxides, quartz, biotite, pyrite, kyanite, and calcite. There is a distinct fining of grains from the vein center, where grains average ∼10 µm, toward the contact with the wall rock (∼7 µm), which probably reflects chilling along the margin (Figs. 4 and 5C). Wall-rock fragments within the matrix (Fig. 5C) generally are angular and flattened and range from 3 mm to 10 µm.

Garnets of the pseudotachylyte matrix occur in three habits: smaller (averaging 10 µm), euhedral ones; larger (up to 100 µm), severely embayed to near-sieve-textured ones; and dendrite- and cauliflower-shaped ones (Figs. 5D and 5E). Embayed garnets have sieve-like textures with bleb-shaped inclusions that include low-Na clinopyroxene, amphibole, kyanite, quartz, ilmenite (Fe-Ti oxide), and calcite. The smaller garnets have fewer inclusions than the larger ones. Dendritic garnets may be hundreds of microns in length and commonly follow linear traces (Fig. 5E). One dendrite trace was observed to terminate at a high angle upon intersecting another trace (Fig. 5E). Other dendrites are more bulbous with cauliflower shapes. Measured garnet compositional ranges are Gr25–30, Py15–19, and Al54–58, but this represents only nine spot analyses. Garnet generally is not present in the gabbronorite host rock except for centimeter-thick zones that parallel the margin of the pseudotachylyte vein (Fig. 5C). In these zones, garnet, and associated biotite, clinopyroxene, and amphibole, occur only where hypersthene grains are cut by the vein. As is characteristic of the eclogite-facies pseudotachylytes of the Bergen Arcs (Austrheim and Boundy, 1994), fluids attending eclogitization do not appear to have penetrated more than a centimeter or two into the dry granulite host.

Plagioclase grains typically range from 10 to 30 µm and show no preferred size or shape. As with other minerals, grain boundaries are often an irregular, polygonal shape, but rounded edges are present as well. Quartz inclusions are common and typically are angular instead of the rounded “bleb” shape of other inclusions. Plagioclase in wall-rock fragments is mostly labradorite (An50–65), whereas plagioclase in the pseudotachylyte matrix is oligoclase/andesine (An20–36).

Clinopyroxene ranges in size from ∼1 to 10 µm and has no preferred shape. It usually is found in contact with plagioclase, amphibole, orthopyroxene, garnet, and opaque grains. Distinct elliptical clusters (roughly 100 µm wide) of mixed plagioclase and low-Na clinopyroxene are interpreted as replaced omphacite grains (Fig. 5F). Compositions of the clinopyroxenes are mainly diopsidic ranging upward with Na content to Jd6.

Amphibole of the matrix ranges in size from 5 to 15 µm and has grain shapes that vary over a wide range. Many grains show cleavage intersections at roughly 60° and 120°, have an elongate shape, and are anhedral. Inclusions of amphibole in other minerals display no preferred shape. Amphibole occurs in contact with all other minerals present. Grain boundaries are usually straight or are a series of short, straight segments defining a curve, but curved segments are also observed. Minerals were identified as amphibole based on shape, BSE color intensity, and presence of K. Amphibole compositions are mostly ferroan pargasite.

Fe oxides and Fe-Ti oxides occur as small (<10 µm), rounded grains in the groundmass, as inclusions in many phases, and also as veinlets. Typically, they stand out as minute bright dots in the BSE images (Figs. 5E and 5F). Some grains display halves of varying Fe and Ti percentages. Hematite occurs in thin (<3 µm wide), <50-µm-long veinlets cutting only plagioclase grains. The veinlets do not appear to follow any crystallographic anisotropies (e.g., cleavage) within the plagioclase grains.

Quartz usually is ∼5 µm and has no preferred shape. It appears to fill voids left by other minerals.

Biotite is rare and usually occurs in aggregates of 2–4 grains and in association with amphibole and plagioclase. Grains range from 10 to 15 µm in length and have no preferred orientation.

Orthopyroxene ranges in size from ∼5 to 15 µm and has no preferred shape. It usually is found incompletely bounded by amphibole (1/2–2/3 of grain boundary), but not as a core with a typical bull's-eye pattern of retrogression. These are also usually found in association with plagioclase. Compositions of the orthopyroxenes range from En61 to En64.

The presence of amphibole and very minor amounts of biotite suggests only minor amounts of fluids during pseudotachylyte formation and/or amphibolite-facies retrogression. It is noteworthy that hydrated minerals are not present within the granulite-facies gabbronorite, only several centimeters outside of the margins of the eclogite shear zones and pseudotachylytes, which is consistent with hydrous eclogitization of the metastable granulites.

The mineralogy and textures preserved in the pseudotachylytes indicate that the primary assemblage that had crystallized directly from the frictional melt was strongly modified by amphibolite-facies retrogression. This should be expected since field relations indicate that the pseudotachylytes formed concomitantly with the eclogite shear zones, which were also highly retrograded. Metamorphic development and retrogression of the shear-zone eclogites are well documented by the work of Kullerud (1992, 1996, 2000) and Markl and Bucher (1997). Although the pseudotachylytes formed through frictional melting of the same host rock, eclogite from the shear zones is coarser grained, preserves relic eclogite-facies minerals (e.g., omphacite and garnet), and is more variably retrograded. Mineral compositional ranges for our pseudotachylyte samples (garnet: Gr25–30, Py15–19, Al55–58; plagioclase: An20–36; clinopyroxene: Jd6) overlap with those reported for the Nusfjord shear-zone eclogites (garnet: Gr10–38, Py10–34, Al50–68; plagioclase: An8–27; clinopyroxene: Jd6–38: from Kullerud, 1992; Markl and Bucher, 1997). Figure 8 plots Kullerud's (1992, 1996, 2000) and Markl and Bucher's (1997) mineral chemical analyses for omphacite and retrograded omphacite, effectively tracking the eclogite- to amphibolite-facies retrogression of the Nusfjord shear-zone eclogites. Our clinopyroxene analyses clearly overlap the most highly retrograded shear-zone eclogite samples (Fig. 8). Similar types of comparisons using garnet and amphibole chemistry (not shown) also indicate that the pseudotachylytes overlap the compositions of the most highly retrograded shear-zone samples.

Despite the aphanitic nature of the pseudotachylyte matrix, several textural features are reminiscent of those reported for the Lofoten shear-zone eclogites. Most notably, the shear-zone eclogites locally preserve omphacite that has been variably replaced, with textures ranging from relic hosts with symplectite rims of low-Na clinopyroxene and albite/andesine (±amphibole) to complete replacement by the same minerals (Kullerud, 1992; Markl and Bucher, 1997; Kassos et al., 2004). Markl and Bucher (1997, p. 20) reported that even where omphacite is not preserved, the mere presence of such symplectites is an “unequivocal indicator that the rock passed through the eclogite stage.” Similarly, we interpret distinct clusters of low-Na clinopyroxene (Jd6) and oligoclase/andesine (An20–36) in the matrix of the pseudotachylytes to be retro-eclogite indicators (Fig. 5F). The extremely fine-grain size (<10 µm) of these matrix minerals is an order of magnitude finer than even the smallest symplectite grains in the coarser eclogites. Although we did not find relic omphacite in our highly retrograded aphanitic rocks, we believe future studies likely will.

Several of our Lofoten pseudotachylyte samples preserve evidence for crystallization directly from the frictional melt. Dendrite garnets from the Lofoten pseudotachylytes (Fig. 5E) are similar to those reported from the Bergen Arcs and Ålesund (Austrheim et al., 1996; Lund and Austrheim, 2003) in their appearance, size (both ∼100 µm, but longer along linear traces), composition (averaging Gr12, Py29, and Al55), and inclusion relations (e.g., orthopyroxene, Na-Ca clinopyroxene, kyanite, amphibole, and dolomite). Austrheim et al. (1996) and Lund and Austrheim (2003) argued that the dendrites reflect rapid (in terms of tens of seconds) crystallization from the melt. The high-pressure inclusions within the dendrites are interpreted to reflect rapid solid-state disequilibrium growth following eclogite-facies seismic failure and pseudotachylyte formation. A lack of equilibrium in our samples is also indicated by the coexistence of several varieties of pyroxenes (Ca-Na pyroxene and hypersthene) and the ranges of compositions of various other mineral phases. Austrheim et al. (1996) interpreted similar mineralogical irregularities of the Holsenøy pseudotachylytes, which are not as strongly retrograded as ours, to reflect rapid disequilibrium growth from the frictional melt. Finally, rapid crystallization from a melt is also indicated by the distinct fining of grains toward the margins of some of the Lofoten pseudotachylyte veins (Fig. 5C), a feature also seen in the Holsenøy eclogite-facies pseudotachylytes (Austrheim and Boundy, 1994).

The high degree of retrogression and disequilibrium in the pseudotachylyte samples that we probed did not allow us to confidently assess pressure and temperature conditions of eclogitization. Pressure-temperature estimates determined for eclogitization within the cogenetic shear zones are >∼1.5 gPa and ∼680 °C (minimum estimates for Flakstadøy eclogites from Markl and Bucher, 1997). Assuming reasonable bulk-rock densities, this pressure estimate suggests >∼45 km depth for shear-zone eclogitization. Frictional melting, which is a very high-strain rate phenomenon (>10−1 s−1; McKenzie and Brune, 1972; Sibson, 1975; Spray, 1995), likely occurred over a time frame of less than a few tens of seconds, whereas the plastic shears could have formed over millions of years. The pseudotachylytes could have formed at nearly any time during operation of the eclogite shear zones (see below) at crustal levels >∼45 km but well below the ∼30 km paleodepth estimated for the amphibolite-facies retrogression (Hodges et al., 1982; Kullerud, 1992; Mooney, 1997).

The Flakstadøy eclogite-facies pseudotachylytes and shear zones are similar enough to those of the Bergen Arcs and Ålesund to suspect a common mechanism for their development. Early interpretations of the Bergen Arcs pseudotachylytes were that fluid-driven eclogitization of the anhydrous granulites resulted in the observed ∼10% volume decrease, providing a causative link between eclogitization and deep-crustal (<60 km) seismic failure (e.g., Pennington, 1983; Hurukawa and Imoto, 1992, 1993; Austrheim and Boundy, 1994). Pseudotachylyte formation, however, requiress hearing along fractures and/or fault planes (i.e., mode II fracturing: McKenzie and Brune, 1972; Sibson, 1975). Later workers, therefore, stressed the change in rheology and the fact that the dry, rigid, nonreacted granulites accommodated flow in the evolving, fluid-mediated, plastic eclogite shear zones by brittle seismogenic failure (Bjørnerud et al., 2002; Lund and Austrheim, 2003; Lund et al., 2004). This interpretation found support from studies on eclogite shears and pseudo tachylytes in the Ålesund area, where, in addition, eclogite-facies hydrofractures are reported (Lund and Austrheim, 2003; Lund et al., 2004).

Any interpretive model for formation of the Flakstadøy pseudotachylytes must accommodate each of the following: crystal-brittle, seismic failure, and frictional melting of metastable granulite; rapid quenching of the melts; crystal-plastic, aseismic flow; and all of these happening together, temporally and spatially, under high-pressure, high-temperature (eclogite-facies) conditions within the lower continental crust. As has already been established on mineral chemical and petrological grounds (Kullerud, 1992, 1996; Markl and Bucher, 1997), our field and structural observations further substantiate that fluid-mediated eclogitization was responsible for the formation of the Flakstadøy shear zones. The pseudotachylytes and cataclasites are preserved at Nusfjord and Skagen because plastic shear strains along individual zones appear to be small and die out only a few centimeters outside of the shears where eclogitizing fluids were able to infiltrate (e.g., Fig. 4). The mechanical conundrum of synchronous crystal-plastic flow and frictional melting (that is, onset of the former should prohibit the latter) in our rocks seems best explained as a spatial phenomenon. Clearly, fluids were limited in their ability to infiltrate far into the dry, rigid granulites. Once hydrated, however, the eclogitized volumes of rock were substantially weakened as crystal-plastic flow mechanisms began to operate. It is reasonable, then, that flow in the eclogite shear zones allowed strain to accumulate within the dry granulitic host until its strength was exceeded, resulting in catastrophic brittle failure and pseudotachylyte formation. Thus, our observations from Flakstadøy are compatible with the Bergen Arcs model.

The rare, plastically deformed pseudotachylyte veins on Flakstadøy evoke a chicken-before-the-egg argument, since they demonstrate that, at least locally, brittle paleoseismic failure preceded plastic yielding. This should be expected, however, given the on-again/off-again, repetitive nature of active and historical seismicity, regardless of focal depth. The Flakstadøy eclogite shears and pseudotachylytes should be viewed in the context of such a cyclical system. Steady-state flow in the shear zones operated continually, analogous to a strip recorder, whereas the pseudotachylytes reflect periodic bursts of seismogenic energy. Pseudotachylyte veins likely would be cannibalized as plastic flow progressed, consuming larger and larger volumes of granulite through time. This would explain the progressive disappearance of the veins toward the centers of some of the thicker eclogite shears. Fluid infiltration would also be enhanced by brittle fracturing, further perpetuating operation of the system.

Available timing, structural, and petrological constraints support that rocks presently exposed in Lofoten were rigid and resilient parts of the lower-crustal levels of an ancient continent from ca. 1.8 Ga until the Middle Ordovician. In the Middle Ordovician, subduction to deeper-crustal levels led to cogenetic plastic shearing and localized frictional melting and pseudotachylyte formation under eclogite-facies conditions (depths ∼45 km). Stiff, metastable (nonreacted) granulite accommodated aseismic, steady-state flow in cogenetic, fluid-mediated, eclogite shear zones by brittle seismogenic failure. Pseudotachylyte veins were likely cannibalized as plastic flow progressed to consume larger and larger volumes of granulite through time. The process operated in an on-again and off-again fashion reminiscent of the repetitive nature of active and historical earthquakes. Later in the Middle Ordovician, these deep-crustal rocks were exhumed to middle-crustal levels, where they were retrograded under amphibolite-facies conditions.

Fluid activity in the continental basement rocks of Lofoten was significant in affecting the metastability and mechanical strength of the roots to the ancient Caledonian mountain belt. During the Early Silurian (Scandian phase), synmetamorphic emplacement of the Caledonian allochthons at mid-crustal levels (∼30 km; Hodges et al., 1982; Steltenpohl and Bartley, 1987) resulted in dewatering reactions such that fluids moved downward to weaken the uppermost structural levels of the granulitic basement complex (Bartley, 1982). Our work in Lofoten indicates that during the early Caledonian, fluids also locally hydrated the dry granulitic basement in the deep crust (only 3–4 km beneath the ancient Conrad discontinuity; Fig. 2) and facilitated its brittle (seismic) failure and plastic (aseismic) weakening.

To our knowledge, this is only the third locality recognized where deep-crustal paleoseismic faults have been exhumed and exposed for direct observation. Our work demonstrates that despite a strong retrograde overprint, careful field and petrological studies on pseudotachylytes preserved in the exposed deep-crustal roots of ancient collisional zones can provide important information on processes controlling the mechanical strength of the deep lithosphere and the generation of deep-foci earthquakes, which has direct application for modern continental seismic zones (e.g., Himalayas, see Jackson et al., 2004).

Much more about the rheology and geodynamics of the lower crust is yet to be learned from future studies in Lofoten. The granulites in Lofoten appear to be only partly reacted to eclogite—even less so than in the Bergen Arcs and Ålesund—an important observation in itself, not only for the genesis of the pseudotachylytes but also for geodynamic models. Presently, we are characterizing additional localities on Flakstadøy, Røst, Værøy, and Vestvågøy (Fig. 2), where we have discovered additional retro-eclogites, shear-zone eclogites with associated pseudotachylytes and garnet-coated fractures, kyanite-clinopyroxene plastic shears, and a remarkably voluminous area of eclogite (the Skagen locality is greater than ∼1.6 km2). Another exciting prospect is that Lofoten may be a continuous column through the entire Caledonian lithosphere rather than an allochthonous terrane emplaced upon Baltic basement, as is the case of the Bergen Arcs. Such continuous columns are exceedingly rare and provide our only opportunity to examine the direct products of lower-crustal deformation in the context of its spatial and temporal relation to deformation that had occurred at shallower levels in the same vertical crustal section (Axen et al., 1998; Beaumont et al., 2001; Klepeis et al., 2003). How strain partitions vertically through the entire lithosphere and is transmitted laterally is one of the more pressing questions concerning the evolution of Earth's continents (e.g., McKenzie et al., 2000; Abers et al., 2002). Ongoing work along the Gullesfjorden and Austerfjord shear zones (Fig. 2) is directed toward assessing how the deep-crustal rocks and structures in Lofoten are related spatially and temporally to the overlying middle- and upper-crustal sections.

This research was made possible through a grant from the Research Council of Norway (to Andresen), and a visiting researcher grant (to Steltenpohl) from the Industrial Liason (IL) Fund (Department of Geosciences, University of Oslo). Kassos thanks the Geological Society of America for helping to support this work. Initial field work for this study was supported by a grant from the Norwegian Marshall Fund (to Steltenpohl). Some initial field discoveries were made while supported by the National Science Foundation (NSF grant EAR-9506698 to W.E. Hames and Steltenpohl). We thank H. Stowell and C. Zuluaga, who helped with microprobe analyses at the University of Alabama, and C. Fleisher, who assisted at the University of Georgia. We gratefully thank Håkon Austrheim for many insightful discussions related to this report. We also thankfully acknowledge K. Klepeis, R. Wintsch, and an anonymous reviewer for provocative reviews that greatly improved this report.