Vertical Coupling And Decoupling in the Lithosphere (Geological Society Special Publication No. 227)

and dt vary rapidly across the region (Fig. 1). Plate boundary-parallel anisotropy is not directly observed at any stations. There are consistencies in certain areas, however. In southern California and in the Sierra Nevada in eastern California, there is a consistent east-west fast direction (Polet & Kanamori 2002; Jones & Phinney 1999). On stations directly above the northern San Andreas fault, the average fast direction is at about N60°W, but the fast directions and delay times vary with the incoming shear-wave polarization. This variation has been explained by an upper layer of plate boundary-parallel anisotropy underlain by a lower layer with direction differing by

11

about 45° (Savage & Silver 1993; Silver & Savage 1994; Ozalaybey & Savage 1995) (Fig. 2). When a shear wave travels sequentially through two anisotropic layers with differing symmetry axes, the waves split in the first layer are split again in the next layer. When the waves are split by less than a period, the resultant wave is a combination of the four split waves and returns an apparent delay time and fast direction that depend on the frequency and also on the incoming polarization direction (Silver & Savage 1994; Riimpker & Silver 1998; Saltzer et al. 2000). Extensive searching for the twolayer phenomenon in southern California has yielded evidence of a much smaller fault-parallel layer, with a delay time of 0.2 s or less, that could be accommodated in the crust. Initial explanations for the two layers in northern California suggested that lithospheric strain acted independently from the asthenosphere, or that the lower layer represented the freezing of past flow along the edge of the retreating Gorda slab (Savage & Silver 1993; Ozalaybey & Savage 1995; Hartog & Schwartz 2000). Both these explanations implicitly assume vertical decoupling, either between the lithosphere and the asthenosphere, or between the upper and lower lithosphere. Savage (1999) offered an alternative suggestion that the upper and lower layers could both be explained by the same shear system. In this scenario the upper layer of anisotropy is caused by large shear strain that had rotated olivine <2-axes into the shear plane, while the lower layer represented small shear from the initial stages of strain, with a-axes oriented at 45° to the shear plane (e.g. Zhang & Karato 1995). This latter model also explained the east-west fast anisotropy beneath the Sierran foothills by a small shear strain. Because this model includes shearing parallel to the San Andreas fault in both layers, it requires more coupling than the previous models. Some authors use structural arguments and the shear-wave splitting results to suggest that the strain beneath the San Andreas fault is accommodated mainly in the lithosphere of the continental North American plate rather than on the colder Pacific plate (Teyssier & Tikoff 1998). Finally, Silver & Holt (2001) suggest that the east-west fast region stems from a reasonably large-scale asthenospheric flow. This flow is assumed to be caused by shear between the lithosphere and the asthenosphere. They further invert for the flow velocity, assuming that simple vectorial summing of the absolute plate motion at the surface with a deeper, constant flow field yields the splitting measurements at most of the western US stations. They determine that such a flow is directed eastward at 50+15 mm a"1.

12

M. K. SAVAGECTAL.

2-layer Anisotropy

They suggest that the mantle flow may have been generated by the subduction of the old Farallon plate. This model assumes that the lithosphere does not contribute to the splitting, which is a reversal of previous interpretations (Silver 1996). Although the San Andreas and Alpine faults are often considered comparable, there are some notable differences that might help to explain some of the discrepancies in splitting data. First, the Alpine fault is fairly straight, with similar transpression along its whole length. The San Andreas fault bends, so that there is larger compression in southern California than further north (DeMets & Stein 1990). The Alpine fault separates two regions of continental crust (and presumably continental mantle) (Sutherland 1995; Okaya et al 2002), while the San Andreas fault separates the North American continental plate and the Pacific oceanic plate. Thus the rheologies across the two faults are expected to be quite different. We use models of mantle flow with varying kinematic conditions to examine changes in anisotropy with distance from a strike-slip fault. We examine the effect of viscosity structure and style of deformation, and test the hypothesis that the measurements along the Alpine fault and the San Andreas fault are explained by shearing in the mantle. Features we try to fit are (Fig. 1) as follows. 1

Fig. 2. Fast directions and delay times from synthetic seismograms (solid triangles) as a function of incoming polarization direction, calculated by the method of Keith & Crampin (1977) through a medium with two layers of anisotropy with fast axes separated by 45°, with average splitting properties of -45°, 1.0 s in the upper layer and 90°, 1.4 s in the lower layer. The dark line is the theoretical curve from the single-frequency analysis of Silver and Savage (Silver & Holt 2001). Open circles are data from station BKS in Berkeley, California (Ozalaybey & Savage 1995). Coordinates are geographical, so that -45° is NW and is parallel to the San Andreas fault.

New Zealand: fault-parallel anisotropy over most of the country, with fast directions that are similar to surface strain directions along the Alpine fault except in the central South Island where fast directions are more fault-parallel. Delay times do not change with distance from the Alpine fault. 2 Northern California: ESE-WNW to eastwest fast directions. Fast directions tend to be more fault-parallel (20° counterclockwise [ccw] from the fault) at stations close to the fault, and closer to 45° ccw from faultparallel at stations further from the fault. Observed splitting varies as a function of phase polarization consistent with two-layer anisotropic models at stations close to the San Andreas fault. Delay times do not change with distance from the fault. 3 Southern California: east-west to ESEWNW fast directions almost everywhere. A weak polarization dependence may exist at some stations directly above the San Andreas fault, but if a shallow layer of fault-parallel anisotropy does exist, it is much thinner than that in northern California. Delay times do not change with distance from the fault.

SEISMIC ANISOTROPY AND COUPLING

We will show that to fit the polarization dependence near the northern San Andreas fault, a rapid change of viscosity with depth is needed.

Method We use techniques outlined previously (Fischer et al 2000; Hall et al. 2000) to calculate flow, strain, seismic anisotropy and the resultant shear-wave splitting sensed by shear body-wave phases (see Fig. 3). In the following discussion, by 'flow' we mean the movement of material in a fluid, which is the definition used by the fluid mechanics community as well as the general public. We first use a finite element code to calculate viscous flow (velocity) in the fault-normal plane, using a two-dimensional Eulerian model and neglecting gravity. The velocity field, which is independent of time, is then used to calculate accumulated finite strain along flow lines following the method of McKenzie (1979) and described in detail elsewhere (Hall et al. 2000). The cumulative strain varies with time, and is calculated by following the particle in a Lagrangian model. In the general case, a particle will pass through different velocities and hence

Fig. 3. Model parameters used in this study. The fault is assumed to strike north. Fast directions () are measured clockwise from north. Strike-slip motion of ± U is imposed from the surface to a depth of 15 km, corresponding to the brittle crust. An arbitrary viscosity structure can be placed in the grid, within which a spacing of 2.5, 5 or 10 km is used. As discussed in the text, the 10 km grid was used for the models presented. The model is two-dimensional; while the motion can be in any direction, the viscosity can only vary either vertically or as a function of distance from the fault. The ellipses show schematically how the strain field varies in each block for an isoviscous medium, getting more strained at the top of the model.

13

deformation tensors during its path. We impose velocity boundary conditions at the top and bottom of the plates, and zero-stress (free-slip) boundary conditions on the sides of the model. The model extends 800km in depth and 1600 km in width. Except where noted, the velocity at the bottom of the model is set to zero. This has the effect of adding a vertical shearing component to the imposed, strike-slip, surface boundary conditions. One could impose conditions at depth identical to those at the surface and avoid the vertical shearing. However, our conception of the physical situation is that the mantle should be relatively fixed at depth compared to the plates above. Furthermore, one of our purposes was to test the hypothesis that the decrease in shear with decreasing depth expected for such a situation would lead to the observed dependence of apparent splitting parameters with depth. The boundary conditions we have chosen implicitly assume plate-driven flow from the top, in this case, the crust. Tests with imposed exponential decay of strain as would be expected from plates driven from the sides showed a much more rapid decay of strain with distance away from the fault. To try to minimize the effects of the side and bottom boundary conditions, we perform our calculations over a wider region, both horizontally and with depth, than we are interested in examining. For all of the calculations, we consider the top 15 km of the model to be the brittle crust and to move with the plate boundary conditions applied. We use two-dimensional grids in the fault-normal plane, so the models are invariant in the faultparallel direction, but we allow the directions of flow and strain to vary in three dimensions. Under this condition, the calculation for the fault-parallel velocity field (Ux) is separable from the fault-normal velocity field (Uy in the horizontal direction and Uz in the vertical direction). Finite strain of the viscous mantle, calculated from the Cauchy deformation tensor (Malvern 1969), is updated along particle flow lines at fixed time steps. As discussed in more detail in the following section, two sets of assumptions were made regarding the relationship of anisotropy to strain. In the first case, olivine a-axes are aligned parallel to the direction of maximum finite extension, and in the second, they are parallel to the particle velocity. In general, we could reliably calculate strain evolution for up to 500 Ma before instabilities led strain ellipsoids for adjacent grids to vary widely. Tests showed little differences (up to 3-4° in polarization and 0.05 s in time delay) between the

14

M. K. SAVAGECTAL.

splitting parameters calculated using grid spacings of 2.5-10 km, so we used a 10km grid for most calculations. For time steps ranging from 0.6 Ma to 0.05 Ma, the calculations for the same total strain were up to 5-6° different in orientation for the fast directions, but the delay time differences remained less than 0.05 s. Because individual SKS measurements have error bars of at least 10° and 0.2-0.3 s, we consider these differences negligible. Whole-Earth models of mantle convection have found that viscosity changes should be less than a factor of 3 between volume elements to avoid numerical instabilities (Moresi et aL 1996). However, some more recent work in smaller-scale situations along the San Andreas fault has found that numerical calculations are still stable with changes in viscosity of up to two orders of magnitude (Malservisi et al. 2002). We tested this by initially calculating most models with a 5 km grid and changes of no more than a factor of 3 between each grid point. For this grid the boundaries were at 400km depth and ±400 km from the fault. The results close to the fault were within a few degrees and a few hundredths of a second of those calculated with a 10 km grid and 800 km to the top and sides of the model. Results 200 km from the fault were substantially different due to the closeness of the side boundary. Thus, in most of our models we used the 10 km grid. This allowed changes in viscosity up to a factor of 9 between each volume element, but allowed our measurements to be less influenced by the boundaries at 800km. In addition, because of large changes in viscosity with depth calculated for oceanic plates (Hirth & Kohlstedt 1996), we allowed changes of up to an order of magnitude or two for the model 'WUS + PAC discussed below. Changing depth to isotropy/anisotropy boundary We examine the effect of the depth to the transition from isotropy to anisotropy. The Preliminary Reference Earth Model (Dziewonski & Anderson 1981) includes anisotropy only in the upper 220 km. Many inversions of surface and higher modes, and body-wave data (Nishimura & Forsyth 1989; Montagner & Tanimoto 1991; Gaherty & Jordan 1995; Gaherty et al. 1996; Debayle & Kennett 2000<2, b) suggest that significant radial anisotropy is confined to the upper 200-300 km of the mantle, with deeper anisotropy beneath continents than under ocean basins (Gaherty et al. 1996). On this basis, it

has been suggested that the Lehmann discontinuity at about 220-300 km depth (Lehmann 1959, 1961) may be caused by a transition from dislocation creep and anisotropic material in the upper region to diffusion creep and isotropic material below. In contrast, recent global tomography has provided evidence for localization of radial anisotropy beneath cratons at 250-400 km depth and has been used to suggest that the Lehmann discontinuity represents the transition from lithosphere with at most weak anisotropy to a strongly anisotropic asthenosphere (Gung et al. 2003). Despite this uncertainty in the depth extent of seismic anisotropy, it is reasonable to assume that, in the actively deforming plate boundary regions that we examine, anisotropy extends to shallow depths that include the lithosphere. We allow anisotropy in our models from the base of the crust to two possible depths within the mantle: 200 km or 300 km. Including anisotropy at greater depths would in general produce results similar to those obtained with the maximum depth at 300 km, primarily because strains at depths greater than 300km are typically small. The 15 km thick crust is assumed to be isotropic. To calculate shear-wave splitting, seismic ray paths are traced through the AK135 model (Kennett et al. 1995). For each block through which the ray passes that is above the assumed isotropy/anisotropy boundary, a tensor of elastic coefficients is rotated to the orientation prescribed by the assumed relation between anisotropy and deformation. A linear particle motion with a period of 10 s (sometimes varied to give more stable measurements) is passed through the first block and split according to the block's anisotropy and the wave's propagation direction and polarization. The resultant particle motion is split again in the next block, and so on until the top of the model is reached. The net effects of splitting observed at the surface are measured using the same type of particle motion analysis (Silver & Chan 1991) that is typically applied to real data. We present only vertical incidence waves here, so the velocity model should not affect the ray paths. The method differs from anisotropy calculations by Tommasi and coworkers (Tommasi 1998; Tommasi et al. 1999) and Blackman and coworkers (Blackman et al. 1996; Blackman & Kendall 2002) in that it does not attempt to model directly the microscale development of anisotropy; we model the velocity and strain and assume a relationship between deformation parameters and anisotropy. The strengths and limitations of this approach are discussed in the following section.

SEISMIC ANISOTROPY AND COUPLING

Relation between deformation parameters and anisotropy The lack of a complete understanding of the relation of macroscopic strain to microstructural development of anisotropy is the biggest obstacle in interpreting seismic anisotropy. In the case of coaxial strain, numerous studies (McKenzie 1979; Wenk & Christie 1991; Ribe & Yu 1991; Ribe 1992) show that the a-axes (fast axes) of olivine should align with the maximum finite extension direction, the fr-axes (slow axes) align with the maximum shortening or contraction direction, and the c-axes (intermediate) align with the intermediate principal strain. However, simple shear experiments (Ribe 1989; Zhang & Karato 1995; Bystricky et al 2000) indicate that, at shear strains of larger than unity, dynamic recrystallization allows olivine a-axes to rotate parallel to the shear direction. At very large strains, shear direction and maximum finite extension become parallel and this distinction is lost. Numerical studies that incorporate the effects of dynamic recrystallization using either viscoplastic self-consistent theory (Wenk & Tome 1999) or a kinematic approach (Kaminski & Ribe 2002) have been successful at modelling such apparent evolution of olivine lattice preferred orientation with increased shear strain. It is important to note that lattice preferred orientations in more general flow fields manifest more complex behaviour (Chastel et al 1993; Blackman et al 1996; Tommasi 1998; Tommasi et al. 1999, 2000; Blackman & Kendall 2002; Kaminski & Ribe 2002). In a number of these models, olivine a-axes appear to align with the direction of flow over substantial regions (Blackman et al. 1996; Tommasi 1998; Tommasi et al. 2000). According to Blackman & Kendall (2002) and Kaminski & Ribe (2002), this result is expected for model regions where the change in flow is substantially slower than the development of preferred orientation and a-axes align parallel to the direction of finite extension for infinite strain, although at small strains the a-axes will follow the maximum extension direction of the strain ellipse for the actual amount of finite strain. On the other hand, Blackman & Kendall (2002) conclude that, over large regions of the flow fields they examine, the fast directions of P-wave anisotropy, which may be considered a rough proxy for olivine a-axes, align parallel to the actual finite extension direction. Finally, simpleshear experiments using water-saturated dunites suggest that the pattern of lattice preferred orientation may differ dramatically from that expected for relatively dry mantle (Jung & Karato 2001),

15

possibly due to a change in the activity of olivine slip systems (Kaminski 2002). In this pattern, the c-axes orient parallel to the finite extension direction. In this study we test two end-member forward models, 'flow-controlled' anisotropy in which olivine a-axes align with the local flow line (i.e. with the particle velocity field) and 'finite strain-controlled' anisotropy in which olivine a-axes align parallel to the direction of maximum finite extension. We do not explicitly consider variations in olivine lattice preferred orientation (LPO) related to high water content. Although a more accurate determination might be made by including variations in lattice preferred orientation development that depend on strain, the simple models used here provide a first-order sense of which flow fields fit the broad features of the observed splitting. Furthermore, because of existing uncertainties in lattice preferred orientation development with respect to strain, as well as other uncertainties in model parameters, more complicated models of lattice preferred orientation development are not obviously necessary. In most of the strain-controlled cases, we consider alignment of orthorhombic olivine rather than the approximation of transverse anisotropy that is assumed in many interpretations of shear-wave splitting (e.g. Savage 1999). We orient the a-axes with the maximum finite extension, &-axes with maximum shortening and c-axes with the intermediate strain. There is a wide variation between the percentage anisotropy measured in different mantle samples (Ismail & Mainprice 1998). Since a definitive quantitative relationship between strain and percent anisotropy is not presently accepted, we place less emphasis on the absolute delay times in our models. We concentrate instead on the pattern of delay time and polarization changes across the fault and with incoming polarization. We use two different cases in different models. For the strain-controlled models, which can be represented by orthorhombic symmetry, we use an average mantle sample built using the compositions and LPO of five xenoliths from South African kimberlites (Mainprice & Silver 1993). We use an increasing delay time (df) for increasing strains according to the following equation:

16

M. K. SAVAGE£TAL.

where p is the path length through the block, Vmin and Vmax are the minimum and maximum velocities for the S-wave velocities for the particular azimuth and angle of incidence of the ray for the (Mainprice & Silver 1993) coefficients, a and b are parameters that can be varied, and 12 and /3 are the middle and smallest eigenvalues of the deformation tensor, respectively. The parameter a controls how rapidly the delay times reach saturation for high strain, and the parameter b controls the maximum magnitude of the anisotropy. We use a = 2.5 and b = 0.2 so that the shape of the shear strain v. delay time curve approximates that presented in Tommasi (1998), and so that the maximum splitting delay times are around 2.0s for model run times of 11.8 Ma. This corresponds to a maximum velocity anisotropy of about 18%, which is rather high. For the flow-controlled models (and one test strain-controlled model), which interpret the orientation of the olivine a-axes as parallel to the direction of flow but do not distinguish the other directions, we use olivine with a-axes oriented in the flow (or maximum finite extension for the case of the test strain-controlled model (AFR2)) direction and equal mixtures of b- and oaxes in the orthogonal planes, with a constant delay time independent of flow velocity (or strain for the case of the test model (AFR2)). We again scale the delay times so that the maximum splitting is about 2 s. This was accomplished by using a value of I? of 1.0 and by substituting the eigenvalue ratios in the strain relationship by the number 2.5. Because of this rather ad hoc scaling, we do not place importance on the absolute delay times and only assess approximate fast direction trends (AFR2).

Results for relative plate motion with an isoviscous model To test the strain-controlled model for the strikeslip motion, we compared the calculated strain from the velocity field for an isoviscous model with that for the analytic solution. The solution for the velocity u at infinite time in a constantviscosity half-space with plate motion of +U and — U imposed at the surface on either side of a fault, if the displacement goes to zero at an infinite distance from the surface, is

The coordinate system notation is given in Fig. 3. The strain rate tensor calculated from this velocity field has eigenvalues of 0 and

Taking the smallest eigenvalue as the negative root above leads to its eigenvector as the minimum principal strain rate. The eigenvector is

The azimuth and plunge of the eigenvector are thus

and

Modelling results In the first flow models we consider, viscosity is assumed to be uniform. These isoviscous models allow us to test our numerical flow calculations against an analytic flow solution, to better understand how strain evolves spatially and with time, and to assess the impact of boundary conditions and other modelling assumptions on our results. We then explore the development of anisotropy in a series of models with variable viscosity. In this section we only briefly compare the results of the isoviscous and variable viscosity models to the overall features of observed splitting across the Alpine and San Andreas faults. A more detailed comparison of the predicted and observed splitting is deferred to the Discussion section.

respectively. These values are for the infinitesimal strain. For such a strain, the eigenvalues of the deformation tensor will all be very close to unity. For infinite strain, the eigenvector progressively rotates, such that the maximum extension (assumed parallel to the olivine a-axis) is along the fault strike, the minimum extension (olivine Z?-axis) is in the — $ direction, and the intermediate extension (olivine oaxis) aligns as radial 'spokes' away from the origin at the fault boundary (Fig. 4). The line at the surface where the two plates meet is a singularity. Because of this singularity, we consider stations slightly off the boundary, at least 2 km distance. The paths passing near the singularity are not affected too much in the final splitting measurement,

SEISMIC ANISOTROPY AND COUPLING

17

Fig. 4. Velocity field for the isoviscous solution, represented by grey shades. The analytic solution contours are represented by the dashed lines. The curvature in the numerical solution results from the zero-stress boundary condition at the edge of the model, which requires the lines of constant velocity to be perpendicular to the boundaries.

because the splitting is an integrated effect over the whole path and the singular region is small. Figure 5a shows the azimuth and plunges of the maximum extension direction as a function of depth and distance from the fault, for an isoviscous model run for two steps or 0.1 Ma (3.4 km), compared to the analytic solution just described. The model includes the following boundary conditions: at the top of the plate, the velocity is +17 mm a"1 for all points to the right of the fault and —17 mm a"1 for all points to the left. The bottom boundary has zero velocity at all points. This isotropic, relative plate motion is called 'Isorel' (Table 1). These conditions approximate the strike-slip components of both the San Andreas fault (SAP) and the Alpine fault (AF) (DeMets et al 1990; Walcott 1998). The azimuths and plunges of the analytic and calculated solutions are within a few degrees of each other at all depths and all distances. A slight mismatch at 197 km and near the surface at 57 km is due to the effects of the side boundary and of including the 15 km crust in the numerical model. Figure 5b and c show the azimuths, plunges and eigenvalues of the maximum extension direction for the isoviscous model run for 11.8 Ma (400 km). Note that the eigenvalues here are from the deformation tensors, and have values of 1 for zero strain. Compared to the infinitesimal case, the fast directions are more fault-parallel, particularly towards the surface, and the plunges are more horizontal. The largest strains are near the surface, particularly for the region closest to the fault. At all depths, the region closest to the fault has plunges closer to 0° and azimuths

closer to 135° (—45°) compared to the more distant regions. At infinite strain, and in the flow solution, all regions show similar behaviour: fault-parallel azimuths and horizontal plunges. To compare the splitting from the isoviscous model with that expected for the analytic solution, we use two different approaches to calculate the velocity fields that are input to the strain code. For the 'calculated' solution, we use the isoviscous model just described. For the 'analytic' solution, we use equation (1) above for the velocity field. In both approaches, the strain is calculated numerically for elapsed times ranging from 6 Ma (200 km displacement), through to 235 Ma, representing an unrealistic 8000 km of displacement that is meant to approach the limit for infinite time. The splitting is also calculated numerically, yielding slight variations in the solutions. The 'analytic' and 'calculated' values differ most for small strains in the centre of the model, but are still within 10°, which is the usual error bar observed in shear-wave splitting results (Fig. 6). The delay time differences between the analytic and calculated solutions differ by less than 0.3 s throughout. The edges of the model are affected by the boundary conditions of zero stress. We calculate the results for a model extending 800 km on either side of the fault and only consider the results within 200 km of the centre of the fault. The high delay times for large strain are due to the high anisotropy used and to the increasing strain at deeper levels with time. At 11.8 Ma strain (which we use as shorthand for strain developed over 11.8 Ma) we have included a calculation based on using the

18

M. K. SAVAGE ETAL.

hexagonal anisotropy model used for the flow calculations. In this model each grid has the fast direction oriented in the maximum strain direction, but has its delay time magnitude independent of strain. It is labelled 'const, delay' in the plots (Fig. 6 and subsequently).

The splitting parameters (Fig. 6) include the integrated effect of the strain up through the model calculated for a 200 km deep anisotropic layer (Fig. 5b & c). Results for a 300 km deep layer are nearly the same, due to the low strain between 200 and 300 km depth. Except for the

Fig. 5. (a) Plunges and azimuths of the maximum extension direction for the case of two steps or 0.1 Ma (3.4 km) of strain. The analytic solution given in the text is shown as solid lines for the plunges and dashed lines for the azimuths. Note the close agreement for the analytic and isoviscous solutions, (b) Plunges and azimuths of the maximum extension direction for 11.8 Ma (400 km) of strain. The fast directions have rotated to become more fault-parallel and more horizontal near the surface, (c) Square root of the ratio of the middle to the smallest eigenvalue of the deformation tensor, which is a measure of the amount of strain. The most highly strained region is that closest to the fault and nearest the surface, (d) Azimuths and plunges of the direction of maximum extension, as in (b), for a strain of 11.8 Ma (400 km) and the WUS + SAP model, (e) Eigenvalue ratios as in (d). (f) Azimuths and plunges for the WUS + PAC model for a strain of 11.8 Ma (400 km), (g) Azimuths and plunges for the Oneside model at 11.8 Ma. (h) Eigenvalue ratios for the Oneside model, (i) Azimuths and plunges for the isotropic APM model, (j) Eigenvalue ratios for the isotropic APM model. (k) Azimuths and plunges for the Decouple model. (1) Eigenvalue ratio for the Decouple model.

SEISMIC ANISOTROPY AND COUPLING

19

Fig. 5. Continued.

hexagonal model, delay times decrease with distance from the fault trace due to the decreased strain and the relationship assumed between percent strain and crystal alignment. For strains less than or equal to 45 Ma, delay times on the fault trace are at least twice as high as those at 200 km from the fault. As stated above, we do not put much faith in the absolute delay time predictions, and while relative delay time patterns

are likely more significant, they are dependent on assumptions we make about how strain scales to anisotropy. For small and intermediate strains, the fast direction is closer to fault-parallel at distances of 0-20 km from the fault than it is at distances of 100km from the fault, although the differences are within the 10° expected errors. This occurs because the larger near-surface strain

Table 1. Models presented Model name

Isorel

Boundary conditions

Top: ±17 mm a"1 across fault Bottom: fixed

Viscosity structure Depth to isotropy/ anisotropy boundary

Figures using model

Notes on fit for finitestrain model

Notes on fit for flow-controlled model

S. SAP and S. AF: fit v. fault distance for 6-8 Ma Central AF: fit only with more rapid olivine alignment Decrease in dt away from fault in model does not fit data N. SAP average fast directions fit, but 4>, dt v. polarization do not fit N.SAF,S.SAFandNZ not well fitted S. SAP and S. AF: fit 0 v. fault distance for 3 Ma Central AF: fit for 800km displacement (23.5 Ma) Decrease in dt away from fault in model does not fit data N. SAP average fast directions fit, but 0, dt v. polarization do not fit Fit to SAP 0 v. fault distance reasonable; , dt v. polarization not fitted

Only central AF fitted Constant dt with distance from fault fits AF

Isoviscous

200 km & 300 km

4, 5a-c, 6, 7

200 km & 300 km

5d, e, 8a-c

WUS + SAP

As above

See Table 2

WUS only

As above

As above but with no weak zone

200km

8d, e

Oneside

As above

5 x 102U everywhere, except a factor of 81 lower from 0-80 km on right

200 km & 300 km

5g, h, 9a, c, d

Same as above Same as above

Same as above

WUS + PAC

As above

See Table 3

200km

5f, 9b, c, d

APM iso

Bottom: fixed Top: Fault-parallel: left (Pacific) side — 46 mm a"1, right (N.A.) 1 mm a"1 Fault-perpendicular: left 9.7 mm a"1, right —19A mm a"1

Isoviscous

200 km & 300 km

5i, j, 10, 11

APM WUS + SAP

As above

In Table 2

200 km & 300 km

11

APM WUS + PAC

As above

In Table 3

200 km & 300 km

11

AsthenEast

Bottom: 50 mm a 1 due E (—35 mm a"1 fault-parallel and +35 mm a"1 faultperpendicular on both sides) Top: Fault-parallel motion on left: — 11 mm a"1, right: +36 mm a""1 Across-strike component left: — 25.3 mm a"1, right: — 54.4 mm a"1 Bottom: Fault-parallel: 0 everywhere Faultnormal: +22 mm a"1 everywhere Top: Fault-parallel: left -57 mm a"1, right: —7 mm a"1 Fault-perpendicular: 0 everywhere

Isoviscous,

200 km & 300 km

Not shown

See Table 4

200km

Bot22mm

5k, 1, 12

Fit to SAP v. fault distance similar to WUS + SAP (f> gets more faultparallel further from the fault. If delay time does not depend on strain, does not change with distance from the fault - does not match data (/> is nearly faultperpendicular from 50-200 km from the fault - not a good fit Same as above

No viscosity models tested fit data well

(/>, dt v. polarization best fitted to twolayer model, cj) does not show quite as strong decrease with fault distance as data

Same as above 4> does not match measurements anywhere

approaches faultperpendicular at 200 km from fault. Does not fit well Flow at 200 km has reasonable fit to SAP 0 v. fault distance No viscosity models tested fit data well

Does not fit at all

22

M. K. SAVAGECTAL.

Fig. 6. Evolution of shear-wave splitting parameters over time, as a function of distance from the fault, calculated for 200 km to the isotropy/anisotropy boundary. Solid symbols with solid lines represent anisotropy calculations based on the numerical strain calculation from the analytic flow field. Open symbols with dashed lines represent corresponding anisotropy based on the finite element calculation of the flow field with a uniform viscosity. The solutions are symmetric across the fault, (a) Fast direction (f>. Values for dt < 0.4 s are not shown because their solution becomes unstable in the numerical codes, (b) Delay times also approximate the analytic solution. Symbols are same as in (a).

close to the fault (Fig. 5c) results in greater weight on the upper, fault-parallel layers. For the case of flow-controlled anisotropy, splitting parameters are constant at all distances across the fault and do not evolve with time. Fast directions line up with the shear direction, as shown by the line at 0 degrees for 'infinite time' in Fig. 6a. The constant delay times are caused by the constant anisotropy used in relating flow velocity to anisotropy. For the 11.8 Ma constant delay time-strain case, the results are similar to the variable delay time calculation near the fault. However, the delay times become larger and the fast directions become more fault-parallel further away

from the fault. This is due both to the rotation of the a-axes toward fault-parallel and to the decrease in plunge of the a-axes away from the fault, as shown in Fig. 5b. The independence of delay times on strain in these models means that splitting from the deep regions is not downweighted in the net splitting calculation as it is in the other cases. Note that the maximum anisotropy in a hexagonal symmetry based on olivine a-axes rotations is for propagation close to 45° to the vertical (e.g., fig. 4 in Savage 1999). The observation of variations in shear-wave splitting parameters as a function of incoming polarization at certain stations close to the San Andreas fault represent another test for the flow models examined here. Figure 7 shows predicted fast directions and delay times as a function of incoming polarization for the isoviscous models discussed above for a station 5 km from the fault. (Predicted splittings for a more complex viscosity structure are also shown and will be discussed later.) For comparison, theoretical curves for a two-layer anisotropic system and the data from station BKS on the San Andreas fault (Ozalaybey & Savage 1995) are included, rotated into the coordinate system of the model. The smooth rotation in finite strain and/or flow direction in the isoviscous models does not produce a strong variation in predicted fast direction as a function of polarization, although some aspects of the predicted delay times are reminiscent of the observed delay time trends. Because the fast direction predictions are significantly more robust and less dependent on modelling assumptions than the predicted delay times, we conclude that the isoviscous models do not provide a good fit to the observed polarization dependence. Effect of viscosity structure The Earth has a variable viscosity, and in particular it has been suggested that the San Andreas fault is very weak, with low viscosity (e.g. Flesch et al. 2000). Including variable viscosity has the effect of concentrating the strain in lower-viscosity regions (e.g. Turcotte & Schubert 1982). For very large strains in our models, anisotropy is relatively insensitive to viscosity structure because strain ellipsoids approach fault-parallel everywhere. For intermediate strains, the low-viscosity region tends to yield fault-parallel anisotropy and higher strains, and the higher-viscosity region yields fast directions at angles up to 45° to the fault and lower strains. We tested several forward

SEISMIC ANISOTROPY AND COUPLING

23

Fig. 7. Predicted splitting measurements for fast direction and delay time as a function of the polarization of the incoming wave (the back-azimuth in the case of SKS phases). Vertical incidence, stations at 5 km distance from the fault, and time steps equivalent to 11.8 Ma since the inception of San Andreas fault motion (corresponding to 400 km displacement across the fault) are used. Solid lines correspond to the two-layer solution presented in Figure 2 for station BKS at Berkeley, California, rotated so that the coordinate system is the one shown in Figure 3. Circles and stars with error bars are SKS and S measurements, respectively, from station BKS (Ozalaybey & Savage 1995) translated into the coordinate system of the fault. Patterns are periodic with a period of 90° in the incoming polarization direction. Different viscosity structures and anisotropic thicknesses lead to slightly different patterns (see text and Table 2). The delay time variations are similar to those of the BKS data, while the fast directions exhibit less variation in these finite element models than are observed in the data.

models of the San Andreas fault where we assumed simple but variable viscosity structures. Symmetric weak fault model (WUS + SAF). To estimate the viscosity structure near the San Andreas fault (SAF), we started with the vertical viscosity structure of the Western United States (WUS) constrained by lake level rebound in the nearby Basin and Range (Bills et al. 1994) (Table 2). In addition, a region of very low viscosity was imposed along the San Andreas fault, following the results of Flesch et al. (2000), by dividing the viscosity in the model by 27 at all depths for a region within 50 km of the fault (Fig. 8a, Table 2).

The biggest velocity gradients, and hence the biggest strain, are concentrated within the lowviscosity regions (Figs 8 & 5e). The surface strain is particularly concentrated along the fault, caused by the narrow lateral zone of low viscosity. This is exemplified in plots of the eigenvalue ratios and the dips and plunges of the resultant olivine a-axes (Figs 8 & 5d, e), calculated at 11.8 Ma. In this region (out to about 27 km from the fault), the azimuths of the maximum extension, and hence the a-axis azimuths, are close to fault-parallel and their plunges are nearly horizontal. Outside the low-viscosity fault region, the largest strain is at depths of 70-120 km. In general, regions of low strain correspond to

24

M. K. SAVAGE ET AL. Table 2. WUS + SAF viscosity model used Depth range of layer (km)

Viscosity away from fault (Pa s)

15-40 40-70 70-125 125-150 150-300 300-800

5 x 1020 5 x io20-4 x 1017 4 x 1017 4 x 1017-2 x 1020 2 x 1020 1 x 1019

regions with the fast direction at —45° (135°) to the fault with variable plunges, while high-strain regions have a-axes closer to fault-parallel and plunges approaching horizontal. However, the details of the strain field become quite complicated at shallow depths. The fast directions predicted for this model are nearly fault-parallel both far from the fault and close to the fault. But between 0 and 37 km from the fault, however, the fast directions become progressively further from fault-parallel for the 11.8 and 6 Ma cases, reaching —25° at 40 km for 6 Ma (200 km displacement; Fig. 8). The case with hexagonal anisotropy and delay times independent of strain also becomes more fault-parallel further from the fault, similar to what occurred with the above i so viscous case. This can be explained by the behaviour of the azimuths of the maximum extension direction (Fig. 5d). For stations within the lateral bounds of the low-viscosity region, anisotropy in the uppermost regions of the model contribute the most to the predicted cumulative splitting parameters, both inherently (e.g. Saltzer et al.

Viscosity next to fault (Pa s)

1.9 1.9 4.8 4.8 9.5 4.8

x x x x x x

1019 1019-4.8 x 1015 1015 1015-9.5 x 1018 1018 1017

Lateral distance to fault boundary (km)

50 50 50 50 50 50

2000), and also because strains are largest at shallow depths. The obvious exception is the test case where anisotropy is not scaled with strain. In the low-viscosity region, the azimuths change from —45° at depths of 150km to fault-parallel at the surface, but because strains are larger at or near the surface, the resulting fast directions lie close to fault-parallel. Past 57 km from the fault, the strains are largest between 60 and 120km depth, where the azimuths are fault-parallel, leading to fast directions within 5° of fault-parallel. The delay times are largest in the low-viscosity region near the fault because the strains are largest there. Even though there is a sharp change in fast direction with depth, the resulting splitting parameters do not show a strong two-layer effect in the fast directions, largely because the strain favours the near-surf ace layers so much that they contribute more to the splitting and do not yield two-layer type behaviour (Fig. 7). For comparison, we also show a case for 23.5 Ma of strain (800 km displacement) with the WUS viscosity but no weak zone near the

Fig. 8. (a) Contours of the velocity field as in Figure 4 for the WUS + SAP viscosity model of the San Andreas fault region (Table 2). (b,c) Time evolution of shear-wave splitting predicted for the WUS + SAP model, as in Figure 6. (d,e) Time evolution of shear-wave splitting predicted for the WUS model without a weak zone near the fault.

SEISMIC ANISOTROPY AND COUPLING

25

Time Evolution WUS+SAF Model 300 km to anisot/isot

Fig. 8. Continued.

fault (Fig. 8d, e). It is similar to the isoviscous case and the weak-fault case just discussed. However, in the strain-controlled case, the fast directions near the fault are closer to faultparallel than for the same time evolution as in the other two models. The case where delay times are independent of strain yields similar delay times and fast directions to the corresponding case in the other two models.

Asymmetric viscosity models. The assumption of symmetric viscosity structures across a fault is unlikely to be met in many instances. The oceanic Pacific Plate (PAC) may have higher viscosity in shallow regions than the continental North American Plate because of the weaker materials in continental crust. Teyssier and Tikoff (1998) have suggested that the plate boundary shearing extends into the asthenosphere

26

M. K. SAVAGEETAL.

Fig. 8. Continued.

in a narrow region on the North American side to yield the observed splitting measurements. We have considered two cases of asymmetric viscosity. In the first model (oneside), a uniform viscosity base is reduced by a factor of 81 if it is shallower than 90 km and within 70 km of the fault on the eastern side (Fig. 9, Table 1). The second model (WUS -h PAC) uses the viscosity structure discussed above on the North American

plate (Bills et al. 1994), and an approximation to viscosity profiles for oceanic lithosphere at 20 Ma age (Hirth & Kohlstedt 1996) for a mantle that is depleted of water content due to melting at a mid-ocean ridge (Table 3). The two models both yield velocity fields with velocity gradients, and hence strain, that are largest near the fault (Figs 9 & 5h). The strain in both models is similar to that in the WUS + SAP

SEISMIC ANISOTROPY AND COUPLING

27

Fig. 9. As in Figure 8, but for viscosity structures that differ across the fault. Contours are constant velocity. (a) A simplified viscosity structure, with a base viscosity of 5 x 1020 Pa s. The viscosity is a factor of 81 lower in a block on the right side of the fault, extending from the surface to a depth of 90 km and extending 70 km in width, (b) The material on the right side of the fault uses the viscosity structure determined by Bills et al. (1994) (Table 2) for the western USA, while the material on the left of the fault uses an approximation to the Hirth & Kohlstedt (1996) viscosity profiles (Table 3) to match the Pacific plate parameters, (c) The resultant fast directions on a profile across the model in parts (a) and (b). (d) Corresponding plots of fast direction and delay time as a function of the incoming polarization direction, as in Figure 7, for the Oneside model using an anisotropic thickness of 300 km.

model just discussed: close to the fault (within about 30km), the strain increases rapidly toward the surface, while far from the fault (past about 60 km) the strain either peaks between 120 and 70km depth (WUS + PAC), or is constant near a value of 1 (Oneside). The WUS + PAC model (Fig. 5f) has a very similar pattern of a-axis orientation with depth to the WUS + SAF model (Fig. 5d). The Oneside model shows less dependence on distance, with <2-axis azimuth and plunge becoming closer to fault-parallel and closer to horizontal as the fault is approached (Fig. 5g). On the 'North American' side of the plate boundary,

the models both yield average fast directions that are closer to fault-parallel near the fault (-15° to -8°) and closer to -45° at distances of 50km from the fault (-20° to -12°) (Fig. 9c). However, the WUS + PAC model, like the WUS + SAF model previously discussed, gets closer to fault-parallel at greater distances from the fault. The Oneside model continues to become less fault-parallel out to about 100 km, where it remains almost constant at -27° to -20° to 200km. The delay times (not shown) correspondingly get larger close to the fault, ranging from 0.6 s at 100 km to 3.0 s directly over the fault for both asymmetric

28

M. K. SAVAGECTAL.

Fig. 9. Continued.

models with 300 km depth to isotropy/anisotropy transition. The asymmetric viscosity models also have trouble fitting the variation of delay time and fast direction as a function of polarization observed in California. The best fit is for the Oneside model (Fig. 9d). The predicted delay times are similar to the observed delay time

polarization patterns, although the fit is not exact. However, the fast directions do not match for any of the models. Effect of increasing compression component. We tested a transpressional model by imposing a 10% convergent component to the velocity. Convergence is accommodated by downward

SEISMIC ANISOTROPY AND COUPLING Table 3. Pacific plate viscosity for WUS + PAC model Depth to bottom of layer (km) 40 45 50 55 60 70 800

Viscosity (Pa s) 1 1 1 1 1 1 1

x x x x x x x

1025 1024 1023 1022 1021 1020 1019

flow in the underlying mantle, and the model neglects the effects of crustal uplift or thickening. The major effect of convergence is to impose a more strongly dipping structure to the model, particularly near the fault. Plunging a-axes yield splitting that varies with back azimuth and incidence angle. Furthermore, the fast directions depend more on distance to the fault in this case. Absolute Plate Motion (APM) model - focus on San Andreas fault. As will be apparent in the Discussion, the relative plate motion models fit the New Zealand data reasonably well, while some aspects of the California data are not well fitted. Other flow models have been suggested for California, so we turn to them. Fig. 10 presents the velocity field obtained with an

29

i so viscous model when velocities equivalent to the absolute plate motions (APMs) of the North American and Pacific plates are applied as boundary conditions. The simple analytic model discussed above does not apply to the absolute plate motion case, so we consider only the numerical solution. We fix the bottom boundary at 800 km depth, and the top boundary is moving with the absolute plate motion velocities (DeMets et al. 1990). In the geographical coordinate system, these are 19.4 mm a~ ! at an azimuth of 222° on the North American plate and 47 mm a"1 at an azimuth of 303° on the Pacific plate (Table 1), and the San Andreas fault strikes at about 135°. In our model coordinate system of fault-parallel at 0°, the motion is — 87° on the North American plate and —12° on the Pacific plate. As a result, flow (Fig. 10) is moving mostly to the left beneath the North American plate on the right and mostly into the page below the Pacific plate on the left. Net convergence beneath the plates results in a slight dip in the flow lines beneath the fault, which translates into plunging axes of symmetry. Near the fault, the model olivine a-axes start out at about —45° to the plate boundary at the bottom of the model, and rotate slowly toward faultparallel as they rise (Fig. 5i). The region close to the fault undergoes smaller strain than was present in the relative plate motion cases. This occurs because the three-dimensional nature of the strain yields eigenvalues that are more

Fig. 10. Velocity model as in Figure 4a, but for the APM model discussed in the text with a constant viscosity model. The arrows represent the within-plane component of velocity, with the length proportional to velocity. The two connected lines represent the within-plane projection of the paths along which a particle would travel.

30

M. KSAVAGEETAL.

similar than the simple relative motion case. Further away (>40 km from the fault) the olivine #-axes rotate beyond fault-parallel and in the flow-controlled model far from the boundary the (2-axis azimuths are parallel to the APM. These translate into the splitting fast directions shown in Fig. 11. Figure 11 compares the predicted fast directions using the Absolute Plate Motion model for different viscosity models and for either flow-controlled or strain-controlled anisotropy. The flow-controlled results are no longer the trivial solution of being fault-parallel across the model, and are also no longer coincident with the strain-controlled results for large finite strain. These differences arise because the particles are subjected to different strain regimes when traversing through the model. The integrated splitting for the flow model with the isoviscous structure yields fast directions that are nearly constant at —12° on the Pacific plate side, changing only slowly at positive distances until they are —87° far from the fault on the

North American plate side. The delay times (not shown) are larger closer to the fault. Only the WUS + PAC viscosity structure for flow matches the predicted patterns of fast direction with distance from the fault observed near the northern San Andreas fault. However, the splitting parameters for this model do not match the two-layer pattern observed over the northern San Andreas fault. Other flow models We tested several other models of flow in the asthenosphere, in part motivated by the study of Silver & Holt (2001). In their study, the average horizontal velocity components of a decoupling asthenosphere, assumed equal to splitting fast directions, are summed with plate velocities determined using surface strains from GPS and earthquake focal mechanism information to ultimately obtain the horizontal component of mantle flow underlying the asthenosphere. In the models tested here, velocities

Fig. 11. Fast directions as a function of distance from the fault for the APM model for several different viscosity structures: bottom, isoviscous; middle, WUS + SAP (Table 2); top, WUS + PAC (Table 3).

SEISMIC ANISOTROPY AND COUPLING

at the bottom boundary were set at 50 mm a 1 due east (Silver & Holt 2001). Velocities at the top boundaries were adjusted so that the sum of the top and bottom velocities equalled the absolute plate motion. In this model (AsthenEast), the fault-parallel motion is — 46 mm a"l on the Pacific plate, and +1 mm a 1 on the North Atlantic plate, while the fault-perpendicular motion is 9.7 mm a^ 1 on the Pacific and — 19.4 mm a~ l on the North Atlantic. This model (Table 1, no figure shown) did a poor job of predicting the North American fast directions for all viscosity models examined. When an isoviscous structure was used, the top boundary conditions caused the flow to become nearly fault-parallel at 200 km, the depth at which anisotropy was assumed to begin. When a model with extremely high viscosity from the bottom of the model up to 250 km was used along with the 50 mm a~ l eastward velocity at the bottom of the model proposed by Silver & Holt (2001), the large shear set up between the surface and 250 km depth on the North American plate yielded strongly plunging axes of symmetry close to the fault. However, the Silver & Holt flow model implicitly assumes that all the splitting occurs in the asthenosphere rather than the lithosphere. Consequently, incorporating approximations to their asthenospheric velocities into our modelling that assumes lithospheric deformation may not be an appropriate comparison. The model that provided the best fit to the observed variations of splitting parameters with polarization was an ad hoc model that was explicitly designed to produce two sharp layers of different fast directions close to the fault. In this model (Decouple), the velocity at the bottom was prescribed at +22 mm a"1 in the faultnormal direction (Fig. 12, Table 1). At the top there was zero velocity in the fault-normal direction but in the fault-parallel direction there was — 57 mm a~ l on the Pacific plate and — 7 m m a ~ 1 on the North American plate. We also arranged the viscosity to decouple the surface from the bottom layers (Table 4). We used a surface layer of high viscosity except for a narrow region close to the fault, and a lowviscosity asthenosphere between 70 and 120 km. The flow at the bottom is entirely to the right (translating to NE in the geographical system; Fig. 12a). The flow direction at the isotropy/anisotropy transition is close to the eastward asthenospheric flow determined in Silver & Holt's (2001) analysis. The fault-parallel flow occurs most strongly in the top 100km (Fig. 12). The a-axis alignment changes little with distance from the fault for a strain of 45 Ma (1400km offset; Fig. 5k). The

31

eigenvalues of the deformation tensor show the largest strain at depths of about 100 km and again near the surface (Fig. 51). The fast directions are unstable at small strains for distances greater than 75 km from the fault, because the delay times are so small that the splitting measurements are not well constrained (Fig. 12b, c). The fast directions in the flow model are 45-65° clockwise from fault-parallel. Strains developed over 23-45 Ma are needed to get delay times greater than 1 s away from the fault. The fast directions from the strain model with strain-independent delay times are nearly constant across the fault, at about -40° to fault-parallel. Those for strains developed over 45 Ma show somewhat more change with distance, starting out at about —15° and becoming about —35° at a distance of 200km from the fault. For 23.5 Ma, the fast directions become more fault-parallel with increasing distance from the fault. We get a closer match this time to the two-layer pattern in the fast directions and delay times as a function of incoming polarization, for most distances, for a 45 Ma strain (1400km offset; Fig. 12d). The 700km offset (not shown) is similar, but the fast direction and delay time peaks occur at a smaller incoming polarization direction, so that this model does not fit as well to the data. This final class of model provides the closest response to the northern San Andreas fault results, where a strong dependence of splitting parameters with polarization is observed. Discussion and application of models New Zealand Strike-slip motion has been occurring along the Alpine fault in New Zealand since about 45 Ma, with varying rates that produced 850km of displacement (e.g. Sutherland 1999). While it is obvious that the Earth is not isoviscous, it is instructive to see what elements of the study areas are fitted by such a simple model. Assuming 'strain-controlled' anisotropy, the isoviscous model would yield near-parallel fast directions of about - 15° (Fig. 6) for this displacement at stations directly over the fault, changing little with distance from the fault. When viscosity decreases with depth as in the WUS model, the fast directions rotate towards fault-parallel somewhat more quickly (Fig. 8d). These directions are close to those observed in most of New Zealand. The fast directions in the central Alpine fault are nearly fault-parallel. This can be explained by a viscosity structure similar to the WUS model, or with almost any

32

M. K. SAVAGECTAL.

Fig. 12. (a) As in Figure 10, but for the final model Decouple discussed in the text (Table 1), in which the top and bottom layers are decoupled and velocity conditions are imposed at the top and bottom of the model. (b,c) Evolution of splitting across the fault, as in Figure 6. (d) As in Figure 7, but for the Bot22mm model.

viscosity model if the fast directions align with the flow. This could occur if the strains along the central part of the Alpine fault are large enough to cause fast directions to align everywhere or they are responding to transpression (see discussion below under 'Including compression ...'), or dynamic recrystallization is large (Molnar et al 1999; Little et al 2002). In southernmost New Zealand, where fast directions are not fault-parallel but are instead coherent with the crustal strain extension directions, it appears likely that anisotropy is 'straincontrolled'. An alternative possibility is that the changing dip of the Fiordland subduction zone in southern New Zealand channels flow away from being fault-parallel. This solution is less satisfying because it requires that the similarity between SKS splitting and surface faulting is mere coincidence. However, a caveat to assuming 'strain-controlled' anisotropy in southern New Zealand is that the decrease in delay time with distance from the fault in all the models using orthorhombic anisotropy with straincontrolled delay times does not fit the observed delay times, which are roughly constant with distance from the fault. To fit the uniform delay times with the 'strain-controlled' models, a more rapid saturation of the anisotropy-strain relation is required. When immediate saturation occurs, delay times increase with distance from the fault, as seen in Figures 6 and 8 for the models with constant delay time independent of strain. Thus we expect that an intermediate relationship would fit the data better.

South of the Alpine fault and in southern California, the fast directions are at an angle of about 30° from the fault, consistent with smaller finite strains. In the simple isoviscous model of an infinite fault, the 30° would correspond to between 6 and 12 Ma, or 200-400 km of displacement (Fig. 6), or closer to 3 Ma (100km displacement) if viscosity decreases with depth as it does in the Basin and Range (Fig. 8d). Previous modelling in southern New Zealand (Moore et al. 2002) shows that strain in the southernmost portion may be small because of the termination of the Alpine fault. Although splitting observed in the central Alpine fault may be matched by either 'straincontrolled' or 'flow-controlled' development of olivine alignment, to match the overall pattern of splitting seen across New Zealand, 'straincontrolled' anisotropy is required. Southern California The results in southern California have certain similarities to those in southern New Zealand, namely that the fast directions and delay times are well defined and consistent over a wide area, and are also oriented at about 30° to the nearby fault (Polet & Kanamori 2002). One difference is that the time delays of about 0.75-1.5 s in southern California are about half the magnitude of those in New Zealand. Also, the fast directions in southern California are similar to fast directions in east-central California and to the lower layer that has been

SEISMIC ANISOTROPY AND COUPLING

33

Fig. 12. Continued.

inferred under the northern San Andreas fault in central California, suggesting that they might be part of a wider-scale process. Considerable effort has gone into trying to find if a second, fault-parallel upper layer can be resolved in southern California (Savage & Silver 1993; Helffrich et al 1994; Ozalaybey & Savage 1995; Polet & Kanamori 2002). The most recent results suggest that a somewhat better fit can be achieved by including a very

narrow, thin layer of fault-parallel anisotropy (0.3 s or less) at the surface for a few stations directly over the fault (Polet & Kanamori 2002) than is obtained by a single-layer model. However, the evidence for a clear two-layer splitting variation with incoming polarization is significantly weaker in southern California than in northern California. We therefore do not require a two-layer splitting signal as a test for flow and viscosity models in southern California.

M. K. SAVAGE£rAL.

34

Fig. 12. Continued.

Either 'strain-controlled' or 'flow-controlled' anisotropy could explain the splitting observed in southern California. Assuming 'straincontrolled' anisotropy, the fast directions in southern California match those of the isoviscous flow model (Fig. 6) for 200 km of fault displacement (6 Ma) or the WUS model for 100 km of displacement. In this scenario, the change in plate motions and the reorientation of the San Andreas fault at 6 Ma could have wiped out any existing fault-parallel anisotropy, which was restored only at the surface (and may be controlled by the crust). The present orientations are consistent with only about 3-6 Ma of faultparallel strain. Assuming 'flow-controlled' anisotropy, the east-west-oriented fast directions throughout much of California could be caused by the eastward-oriented flow described by Silver & Holt (2001). The consistency of delay times with distance again requires saturation of the strain-anisotropy relation. Finally, an inferred 'mantle drip' with high P-wave velocities at

depths of 40-150 km in southern California (e.g. Kohler 1999) does not seem to cause any change in delay times or fast directions associated with its location. Including compression in southern California and New Zealand In both southern California and New Zealand, there is about 10° of obliquity, which leads to compression. Including compression in our simple models led to plunging a-axes beneath the fault, which leads to back-azimuthal variations in splitting parameters over the fault and to variations in fast direction with distance from the fault. There is no observational evidence for such plunging <2-axes in either region. Variations in splitting with back-azimuth are negligible in southern California (Polet & Kanamori 2002). The splitting in New Zealand varies little with distance from the fault,

Table 4. Viscosity model used for Decouple model Depth to bottom of layer (km)

50 70 120 800

Viscosity away from fault (Pa s)

10 10 0.3 10

x x x x

1020 1020 1020 1020

Viscosity next to fault (Pas)

0.3 10 0.3 10

x x x x

1020 1020 1020 1020

Lateral distance to fault boundary (km)

30 N/A N/A N/A

SEISMIC ANISOTROPY AND COUPLING

suggesting a variation in plunge of the symmetry axes is unlikely. We can instead consider the results for transpressional shear (plane transpression) (Tommasi et al. 1999). In this case, rather than become rotated to near-vertical, the <2-axes are even more strongly aligned with the fault plane than in the simple shear case. Thus, the results in central NZ could be explained by transpression with this model, but the same model would not explain the results in southern California, where the fast direction is strongly aligned in an east-west direction at 30° from the fault. Northern California Splitting behaviour as a function of distance to the fault. The models for strains of 400 km or 11.8 Ma are expected to match the behaviour in northern California best since they approximate the time that the northern San Andreas fault has been active. Splitting measured on stations directly above the northern San Andreas fault have average fast directions of about —20° to the fault, and delay times of 1.5-2.0 s (Fig. 2). About 200km to the east under the Sierra Nevada foothills, delay times remain large but fast directions are closer to — 45° to the fault. In the southern Sierra Nevada, delay times are smaller and fast directions are again —45° to the fault (Jones & Phinney 2003). None of the models completely match the change in fast directions away from the San Andreas fault in northern California. The models that best match the change are the 'Decoupled' model at 45 Ma duration (Fig. 12b) and the 'Oneside' model (Fig. 9c). For the 'Oneside' model, using 300 km rather than 200 km to the isotropy/anisotropy transition yields a slightly better match to the fast directions, with at -35° at distances of 200km from the fault, and —20° directly over the fault. However, the differences in

35

that comes close to matching the variation in fast directions with back-azimuth at northern San Andreas near-fault stations is the 'Decoupled' model with imposed asthenospheric fault-normal flow at the bottom of the model at 800km, and imposed fault-parallel flow at the surface. This model also provides a reasonable match to the delay times with back-azimuth. It shows less variation in average fast direction with distance from the fault than is strictly observed. However, its predicted fast directions fall within the range of the observed fast directions. The main problem with this model is that it was explicitly designed to match the twolayer splitting observations and we are unaware of any independent evidence for the assumed velocity boundary conditions. In addition, the 45 Ma (1500 km displacement) needed to produce the splitting results is larger than the strains in the region. Such concerns are less central, however, because the strain required will likely be modified with a different viscosity structure. Comparison of northern and southern California The major difference between northern and southern California splitting measurements is the behaviour of stations directly on the fault. On-fault stations in southern California are nearly identical to the off-fault stations, while on-fault stations in northern California have average fast directions that are closer to faultparallel, and strong variations in incoming polarization that fit well to a two-layer model. Although the 'Decoupled' model that best fits the two-layer behaviour in northern California is not entirely satisfactory, it does demonstrate that a low-viscosity channel capable of producing a sharply defined change in olivine alignment with depth can produce the observed variation of splitting with polarization. Therefore, a simple way to explain the lack of clear two-layer splitting in New Zealand and southern California is the absence of such a low-viscosity channel, or perhaps simply that its boundaries are more gradational. Alternatively, our basic assumption that the splitting observed in these regions is due to active deformation may be in error. For example, some of the variations in anisotropy between the regions may reflect 'fossil' anisotropy due to past deformation now frozen in to the lithosphere. One caveat to these arguments is that recent results from nine years of data at station SNZO at the southern end of the North Island suggest that a similar two-layer

36

M. K. SAVAGEErAL.

structure may also occur in New Zealand (Marson-Pidgeon & Savage 2003). In summary, we can explain the fast directions in New Zealand and southern California by strike-slip models with strain-controlled anisotropy, but the lack of variation of delay times with distance from the faults requires a saturation of delay times with strain that occurs more rapidly than predicted with existing models. The smaller delay times in southern California require smaller percent anisotropy or a thinner anisotropic layer. We have used a rather large percent anisotropy (18%) as the maximum, to match the two-layer models on the fault. A smaller maximum value would simply reduce all the delay times to scale. A value of half that used would still provide delay times of the order of magnitude observed in California. But the need for such a high value still suggests that some of our modelling assumptions require modification, or that other factors contribute to the splitting in addition to the strain from plate motions. To fit the apparent two-layer behaviour of the fast directions on near-fault stations in northern California requires a model with substantial decoupling, which we have engineered by inserting two low-viscosity layers separated by a highviscosity layer, and by boundary conditions in which the lower boundary is moving in a different direction from the upper layer. This model also provides a reasonable match to the change in average fast direction with distance from the fault. Conclusions As outlined above, our examination of simple models of strike-slip plate boundaries has allowed us to fit most of the observations in New Zealand and California, but no one model fits all the observations. Model results We have presented a number of models for the shear-wave splitting results expected in regions of strike-slip motion. In particular, we have analysed models that approximate hypotheses previously presented for the San Andreas fault and Alpine fault regions. The following points are of note: 1 Variable-viscosity regions concentrate strain in the low-viscosity regions. Regions with low strain have fast directions close to — 45° from fault-parallel and plunges close to 45°, while high-strain regions have fast

2

3

4

5

directions close to fault-parallel and plunges close to horizontal. Using standard relations between strain and strength of anisotropy limits the contribution of low-strain regions to the integrated splitting through the models. Strain tends to decrease with distance from the modelled strike-slip faults, leading to delay times that decrease with distance from the fault unless the strain-delay time relation saturates at small strains. Flow-controlled relative plate motion models have fast directions lined up parallel to the fault for all viscosity models. For absolute plate motion models, the fast directions in the flow-controlled models are parallel to the APM at large distances from the fault. For the simple relative plate motion models with the viscosity structures examined, we find little difference between results for depths of 200 km or 300 km to the boundary between the anisotropic (dislocation creep) and isotropic (diffusion creep). This is because there is little strain deeper than 200 km. An ad hoc model with a strong low-viscosity layer at depth and with a flow at the bottom of the anisotropy/isotropy boundary that is at about 45° to fault-parallel yields the best two-layer type behaviour of any of the models examined.

Application of models to New Zealand and California 1

Many aspects of the splitting observed in New Zealand and in southern California are reasonably well fitted by the isoviscous models, and also by a stratified viscosity model based on the western USA (Bills et al. 1994). More complex viscosity structures cannot be ruled out, but they are not required. 2 The lack of decrease in observed delay times with distance from the faults in all the regions suggests that there may be a more rapid saturation of the strain-anisotropy relation than that assumed in these models. 3 In the central Alpine fault, NZ, a stratified viscosity structure, large strain, or dynamic recrystallization can all explain the observed fault-parallel fast directions. 4 Both southern New Zealand and southern California have fast splitting directions at an angle of 30° to the faulting, suggesting smaller strain than along the faults further

SEISMIC ANISOTROPY AND COUPLING

5

6

7

north. The strain required to achieve this angle is the equivalent of that expected for about 200-300 km of relative motion on a long continuous fault in an isoviscous medium, or about 100 km for a medium with viscosity like that of the western USA. An ad hoc model, with imposed eastward flow at the base of the anisotropic zone and with a strongly stratified viscosity model, provides a reasonable fit to the variation of fast directions and delay times as a function of incoming polarization in northern California, and provides a reasonable match to the behaviour of the change in fast direction as a function of distance from the fault. This type of model succeeds only if large differences in olivine alignment are engineered in a short vertical distance with rapid viscosity contrasts. If matching the two-layer splitting seen in northern California is not required, then a much wider range of models becomes plausible, including the isoviscous structures possible in southern California and New Zealand. Neither a strict application of the Silver & Holt (2001) model, nor most of the absolute plate motion models, yield fast directions that come close to the behaviour of the fast directions with distance seen in California. The exception is the absolute plate motion model with flow-controlled anisotropy and a 200 km base to the anisotropic transition. This model does not, however, yield twolayer type behaviour over the San Andreas fault. Although the possibility remains that some of the regional differences reflect anisotropy due to past deformation events that is frozen in to a rigid lithosphere, these modelling results show that much of the variation in splitting observed between northern and southern California and New Zealand can be modelled by the interaction of ongoing plate motion with variable-viscosity structures.

We would like to thank Brad Hager, Mark Parmentier, Euan Smith and Sarah Zaranek for help with the analytic solution. Rudy Wenk initially suggested the single-strain model for the two layers of anisotropy. Mark Parmentier, Don Fischer and Kevin Furlong offered many suggestions on modelling and programming. M. Savage was supported for some of this work while on sabbatical at Brown University, with support from NSF grant EAR9903026 and a grant from the NZ/USA Scientific and Technological Co-operative Science Programme (NZ/USA CSP) of the 2000/01 International Science and Technology (ISAT) Linkages Fund. Some of the

37

work was supported by a grant from the New Zealand Marsden Fund.

References AUDOINE, E. 2002. Upper mantle and crustal seismic anisotropy across the Pacific—Australian plate boundary, New Zealand. PhD thesis, Victoria University of Wellington. BILLS, B.C., CURREY, D.R. & MARSHALL, G.A. 1994. Viscosity estimates for the crust and upper mantle from patterns of lacustrine shoreline deformation in the eastern Great Basin. Journal of Geophysical Research, 99, 22059-22086. BLACKMAN, D.K. & KENDALL, J.-M. 2002. Seismic anisotropy in the upper mantle 2. Predictions for current plate boundary flow models. Geochemistry, Geophysics, Geosystems, 3, 8602. (doi:10.1029/ 2001GC000247). BLACKMAN, D.K., KENDALL, J.-M., DAWSON, P.R., WENK, H.-R., BOYCE, D. & MORGAN, J.P. 1996. Teleseismic imaging of subaxial flow at midocean ridges: travel-time effects of anisotropic mineral texture in the mantle. Geophysical Journal International, 127, 415-426. BYSTRICKY, M., KUNZE, K., BURLINI, L. & BURG, J.-P. 2000. High shear strain of olivine aggregates: rheological and seismic consequences. Science, 290, 1564-1567. CHASTEL, Y.B., DAWSON, P.R., WENK, H.-R. & BENNETT, K. 1993. Anisotropic convection with implications for the upper mantle. Journal of Geophysical Research, 98, 17757-17771. DEBAYLE, E. & KENNETT, B.L.N. 20000. Anisotropy in the Australasian upper mantle from Love and Rayleigh waveform inversion. Earth and Planetary Science Letters, 184, 339-351. DEBAYLE, E. & KENNETT, B.L.N. 2000b. The Australian continental upper mantle; structure and deformation inferred from surface waves. Journal of Geophysical Research, 105, 25,423-25,450. DEMETS, C. & STEIN, S. 1990. Present-day kinematics of the Rivera Plate and implications for tectonics in southwestern Mexico. Journal of Geophysical Research, 95, 21931-21948. DEMETS, C., GORDON, R.G., ARGUS, D.F. & STEIN, S. 1990. Current plate motions. Geophysical Journal International, 101, 425-478. DZIEWONSKI, A.M. & ANDERSON, D.L. 1981. Preliminary Reference Earth Model. Physics of the Earth and Planetary Interiors, 25, 297-356. FISCHER, K.M., PARMENTIER, E.M., STINE, A.R. & WOLF, E.R. 2000. Modeling anisotropy and platedriven flow in the Tonga subduction zone back arc. Journal of Geophysical Research, 105, 16181-16192. FLESCH, L.M., HOLT, W.E., HAINES, AJ. & SHEN-TU, B. 2000. Dynamics of the PacificNorth American plate boundary in the western United States. Science, 287, 834-836. GAHERTY, J.B. & JORDAN, T.H. 1995. Lehmann discontinuity as the base of an anisotropic layer beneath continents. Science, 268, 1468-1471.

38

M. K. SAVAGE ET AL.

GAHERTY, J.B., JORDAN, T.H. & GEE, L.S. 1996. Seismic structure of the upper mantle in a central Pacific corridor. Journal of Geophysical Research, 101,22291-22309. GUNG, Y., PANNING, M. & ROMANOWICZ, B. 2003. Global anisotropy and the thickness of continents. Nature, 422, 707-711. HALL, C.E., FISCHER, K.M., PARMENTIER, E.M. & BLACKMAN, D.K. 2000. The influence of plate motions on three-dimensional back arc mantle flow and shear wave splitting. Journal of Geophysical Research, 105, 28009-28033. HARTOG, R. & SCHWARTZ, S.Y. 2000. Subductioninduced strain in the upper mantle east of the Mendocino triple junction, California. Journal of Geophysical Research, 105, 7907-7930. HATHERTON, T. 1968. Through the looking glass: a comparative study of New Zealand and California. Nature, 220, 660-663. HELFFRICH, G., SILVER, P.G. & GIVEN, H. 1994. Shear-wave splitting variation over short spatial scales on continents. Geophysical Journal International, 119,561-573. HIRTH, G. & KOHLSTEDT, D.L. 1996. Water in the oceanic upper mantle: implications for rheology, melt extraction and the evolution of the lithosphere. Earth and Planetary Science Letters, 144, 93-108. ISMAIL, W.B. & MAINPRICE, D. 1998. An olivine fabric database: an overview of upper mantle fabrics and seismic anisotropy. Tectonophysics, 296, 145-157. JONES, C.H. & PHINNEY, R.A. 1999. Prospecting for the petrology of the upper mantle: Teleseismic shear waves in the Sierra Nevada, California. Geological Society of America Abstracts Program, 31, no. 7, A-481. JUNG, H. & KARATO, S.-I. 2001. Water-induced fabric transitions in olivine. Science, 293, 1460—1463. KAMINSKI, E. 2002. The influence of water on the development of lattice preferred orientation in olivine aggregates. Geophysical Research Letters, 29, (doi: 10.1029/2002GL014710), 4 pp. KAMINSKI, E. & RIBE, N.M. 2002. Timescales for the evolution of seismic anisotropy in mantle flow. Geochemistry, Geophysics, Geosystems, 3, (doi: 10.1029/2001GC000222), 17 pp. KEITH, C.M. & CRAMPIN, S. 1977. Seismic body waves in anisotropic media: synthetic seismograms. Geophysical Journal of the Royal Astronomical Society, 49, 225-243. KENNETT, B.L.N., ENGDAHL, E.R. & BULAND, R. 1995. Constraints on seismic velocities in the Earth from traveltimes. Geophysical Journal International, 122, 108-124. KLOSKO, E.R., Wu, F.T., et al, 1999. Upper mantle anisotropy in the New Zealand region. Geophysical Research Letters, 26, 1497-1500. KOHLER, M.D. 1999. Lithospheric deformation beneath the San Gabriel Mountains in the southern California Transverse Ranges. Journal of Geophysical Research, 104, 15025-15041. LEHMANN, I. 1959. Velocities of longitudinal waves in the upper part of the Earth's mantle. Annuales de Geophysique, 15, 93-118.

LEHMANN, I. 1961. S and the structure of the upper mantle. Geophysical Journal of the Royal Astronomical Society, 4, 124-138. LITTLE, T.A., SAVAGE, M.K. & TIKOFF, B. 2002. Relationship between crustal finite strain and seismic anisotropy in the mantle, Pacific-Australia plate boundary zone, South Island, New Zealand. Geophysical Journal International, 151, 106-116. MAINPRICE, D. & SILVER, P.G. 1993. Interpretation of SKS-waves using samples from the subcontinental lithosphere. Physics of the Earth and Planetary Interiors, 78, 257-280. MALSERVISI, R., GANS, C. & FURLONG, K.P. 2002. Numerical modeling of strike-slip creeping faults and implications for the Hayward fault, California. Tectonophysics, 6788, 1-17. MALVERN, L.E. 1969. Introduction to the Mechanics of a Continuous Medium. Prentice-Hall, Englewood Cliffs, NJ. MARSON-PIDGEON, K. & SAVAGE, M. 1997. Frequency-dependent anisotropy in Wellington, New Zealand. Geophysical Research Letters, 24, 3297-3300. MARSON-PIDGEON, K.A. & SAVAGE, M.K. 2004. Modeling shear-wave splitting observations from Wellington, New Zealand. Geophysical Journal International, 157, 853-864. MARSON-PIDGEON, K., SAVAGE, M.K., GLEDHILL, K.R. & STUART, G. 1999. Seismic anisotropy beneath the lower half of the North Island, New Zealand. Journal of Geophysical Research, 104, 20277-20286. MATCHAM, I., SAVAGE, M.K. & GLEDHILL, K.R. 2000. Distribution of seismic anisotropy in the subdue tion zone beneath the Wellington region, New Zealand. Geophysical Journal International, 140, 1-10. MCKENZIE, D. 1979. Finite deformation during fluid flow. Geophysical Journal of the Royal Astronomical Society, 58, 689-715. MOLNAR, P., ANDERSON, H., et al. 1999. Continuous deformation versus faulting through the continental lithosphere of New Zealand. Science, 286, 516-519. MONTAGNER, J.-P. & TAMMOTO, T. 1991. Global upper mantle tomography of seismic velocities and anisotropies. Journal of Geophysical Research, 96,20337-20351. MOORE, M., ENGLAND, P. & PARSONS, B. 2002. Relation between surface velocity field and shearwave splitting in the South Island of New Zealand. Journal of Geophysical Research, 107, 2198 (doi:10.1029/2000JB000093). MORESI, L., ZHONG, S. & GURNIS, M. 1996. The accuracy of finite element solutions of Stokes' flow with strongly varying viscosity. Physics of the Earth and Planetary Interiors, 97, 83-94. NISHIMURA, C.E. & FORSYTH, D.W. 1989. The anisotropic structure of the upper mantle in the Pacific. Geophysical Journal International, 96, 203-229. OKAYA, D.A., HENRYS, S. & STERN, T. 2002. Doublesided onshore-offshore seismic imaging of a plate boundary: 'super-gathers' across South Island, New Zealand. Tectonophysics, 355, 247-263.

SEISMIC ANISOTROPY AND COUPLING OZALAYBEY, S. & SAVAGE, M.K. 1995. Shear-wave splitting beneath western United States in relation to plate tectonics. Journal of Geophysical Research, 100, 18135-18149. POLET, J. & KANAMORI, H. 2002. Anisotropy beneath California: shear wave splitting measurements using a dense broadband array. Geophysical Journal International, 149, 313-327. RIBE, N.M. 1989. Seismic anisotropy and mantle flow. Journal of Geophysical Research, 94, 4213-4223. RIBE, N.M. 1992. On the relation between seismic anisotropy and finite strain. Journal of Geophysical Research, 97, 8737-8747. RIBE, N.M. & Yu, Y. 1991. A theory for plastic deformation and textural evolution of olivine polycrystals. Journal of Geophysical Research, 96, 8325-8335. RUMPKER, G. & SILVER, P.G. 1998. Apparent shearwave splitting parameters in the presence of vertically varying anisotropy. Geophysical Journal International, 135, 790-800. SALTZER, R.L., GAHERTY, J.B. & JORDAN, T.H. 2000. How are vertical shear wave splitting measurements affected by variations in the orientation of azimuthal anisotropy with depth? Geophysical Journal International, 142, 374-390. SAVAGE, M.K. 1999. Seismic anisotropy and mantle deformation: what have we learned from shear wave splitting? Reviews of Geophysics, 37, 65-106. SAVAGE, M.K. 2002. Seismic anisotropy and mantle deformation in the western US and southwestern Canada. International Geology Review, 44, 913-937. SAVAGE, M.K. & SILVER, P.G. 1993. Mantle deformation and tectonics: constraints from seismic anisotropy in western United States. Physics of the Earth and Planetary Interiors, 78, 207-227. SCHERWATH,

M.,

MELHUISH, A.,

STERN, T.

&

MOLNAR, P. 2002. Pn anisotropy and distributed upper mantle deformation associated with a continental transform fault. Geophysical Research Letters, 29, 16-1-16-4. SILVER, P.G. 1996. Seismic anisotropy beneath the continents: probing the depths of geology. Annual Reviews of Earth and Planetary Science, 24, 385-432. SILVER, P.G. & CHAN, W.W. 1991. Shear wave splitting and subcontinental mantle deformation. Journal of Geophysical Research, 96, 1642916454. SILVER, P.G. & HOLT, W.E. 2001. The mantle flow field beneath western North America. Science, 295, 1054-1057. SILVER, P.G. & SAVAGE, M.K. 1994. The interpretation of shear-wave splitting parameters in the presence of two anisotropic layers. Geophysical Journal International, 119, 949-963.

39

SUTHERLAND, R. 1995. The Australia-Pacific boundary and Cenozoic plate motions in the SW Pacific: some constraints from Geosat data. Tectonics, 14, 819-831. SUTHERLAND, R. 1999. Cenozoic bending of New Zealand basement terranes and Alpine Fault displacement: a brief review. NZ Journal of Geology and Geophysics, 42, 295-302. TEYSSIER, C. & TIKOFF, B. 1998. Strike-slip partitioned transpression of the San Andreas fault system: a lithospheric scale approach. Geological Society of London Special Publication, 135, 143-158. TOMMASI, A. 1998. Forward modeling of the development of seismic anisotropy in the upper mantle. Earth and Planetary Science Letters, 160, 1-13. TOMMASI, A., TIKOFF, B. & VAUCHEZ, A. 1999. Upper mantle tectonics: three-dimensional deformation, olivine crystallographic fabrics and seismic properties. Earth and Planetary Science Letters, 168, 173-186. TOMMASI, A., MAINPRICE, D., CANOVA, G. & CHASTEL, Y. 2000. Viscoplastic self-consistent and equilibrium-based modeling of olivine lattice preferred orientations: implications for the upper mantle seismic anisotropy. Journal of Geophysical Research, 105, 7893-7908. TURCOTTE, D.L. & SCHUBERT, G. 1982. Geodynamics: Applications of Continuum Physics to Geological Problems. John Wiley & Sons, New York. VAUCHEZ, A. & NICOLAS, A. 1991. Mountain building: strike-parallel motion and mantle anisotropy. Tectonophysics, 185, 183-201. VINNIK, L.P., MAKEYEVA, L.I., MILEV, A. & USENKO, A.Y. 1992. Global patterns of azimuthal anisotropy and deformations in the continental mantle. Geophysical Journal International, 111, 433—447. WALCOTT, R.I. 1998. Modes of oblique compression: late Cenozoic tectonics of the South Island of New Zealand. Reviews of Geophysics, 36, 1-26. WENK, H.-R. & CHRISTIE, J.M. 1991. Review paper: comments on the interpretation of deformation textures in rocks. Journal of Structural Geology, 13, 1091-1110. WENK, H.-R. & TOME, C.N. 1999. Modeling dynamic recrystallization of olivine aggregates deformed in simple shear. Journal of Geophysical Research, 104,25513-25527. YEATS, R.S. & BERRYMAN, K.R. 1987. South Island, New Zealand, and Transverse Ranges, California: a seismotectonic comparison (USA). Tectonics, 6, 363-376. ZHANG, S. & KARATO, S.-L 1995. Lattice preferred orientation of olivine aggregates deformed in simple shear. Nature, 375, 774-777.

This page intentionally left blank

Mantle-driven deformation of orogenic zones and clutch tectonics BASIL TIKOFF1, RAY RUSSO2, CHRISTIAN TEYSSIER3 & ANDREA TOMMASI4 1 Department of Geology and Geophysics, University of Wisconsin, Madison, WI53706, USA (e-mail: [email protected]) ^Department of Geological Sciences, Northwestern University, Evanston, IL 60208, USA 3 Department of Geology and Geophysics, University of Minnesota, Minneapolis, MN 55455, USA 4 Laboratoire de Tectonophysique, ISTEEM, CNRS / Universite de Montpellier II, 34095 Montpellier cedex 5, France Abstract: Compatible deformation between the upper crust and upper mantle is documented for a variety of ancient and neotectonic settings, suggesting that these lithospheric layers are coupled. Areas of neotectonic deformation are also characterized by high seismic attenuation, indicating that the uppermost mantle is Theologically weak and flowing in these regions. The flow of the mantle, both lithospheric and asthenospheric, potentially drives deformation in continental orogenic zones. Three-dimensional models, controlled by bottom-driven mantle flow, are proposed for obliquely convergent, transcurrent and obliquely divergent plate margins. Our analysis indicates that the absolute, and not just relative, plate motions play a critical role in the orogenic cycle.

The Earth displays a radial compositional stratification, which results in depth-dependent variation in the strength of Earth materials (rheology). The rheology of each of these distinct lithospheric layers was quantified through the use of experimental deformation, and corroborated by geological field studies and microstructure analysis. A steadily accruing set of diverse data indicates that the Earth's continental crust is coarsely divisible into two mechanically distinct layers in continental regions: the upper crust and lower crust. The upper crust deforms by brittle or cataclastic mechanisms, while the lower crust deforms by crystal-plastic mechanism (Sibson 1977; Scholz 1988). The upper mantle underlying continental crust is also mechanically divisible into a more rigid lithospheric upper mantle and a more mobile, fluid-like asthenospheric mantle (Brace & Kohlstedt 1980). Modifications to these rheological estimates have been made, such as the recognition of the importance of micas in upper crustal deformation (Imber et al 2001) or the effect of water in the mantle rheology (Kohlstedt et al 1995). The compositional and rheological heterogeneity of the lithospheric layers was used to infer horizontal decollement zones in the lithosphere, which results in decoupling of surface crustal blocks from underlying mantle (e.g. Oldow et al. 1990). If this scenario is true, it is an

important obstacle for understanding how lithospheric motions and deformation relate to deeper Earth processes such as mantle convection: if there is no mechanical connection, plate tectonics is either independent or, at a minimum, does not have a simple relationship to mantle convection. This basic realization has recently emerged as one of the most fundamental problems of modern tectonics. In this contribution, we synthesize geological, microstructural and geophysical datasets and propose a three-dimensional view of lithospheric deformation. In a variety of settings, upper crustal deformation (characterized by geological mapping and geodetic information) is remarkably similar to mantle deformation (shear-wave splitting data). Consequently, we infer that the layers are at least partially coupled in a process we call clutch tectonics (Tikoff et al. 2002). Alternatively stated, detachment zones are not ubiquitous in the ductile lower crust. Second, we use geophysical arguments to suggest that the uppermost mantle is composed of two distinctly different types: (1) rigid lithospheric mantle under oceanic crust and cratonic regions of continents; and (2) a more fluid, 'asthenospheric-like' mantle under mobile belts. Consequently, we use the term 'upper mantle' rather than lithospheric mantle or asthenospheric mantle, as the rheology of the upper mantle may depend on tectonic setting.

From: GROCOTT, J., MCCAFFREY, K. J. W., TAYLOR, G. & TIKOFF, B. (eds) 2004. Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 41-64. 0305-8719/047$ 15 © The Geological Society of London 2004.

42

B. TIKOFFCTAL.

The concept of plate tectonics has never worked particularly well in mountain belts: plates as lithospheric columns that remain intact through time do not exist in these settings. The observations suggest that crustal deformation is potentially tied in a direct way to underlying mantle movement (e.g. convection), through a series of subhorizontal, high-strain zones which act to partially attach adjacent lithospheric layers. We first present an overview of the observations, both geological and geophysical. We then present the clutch model to explain the observed phenomenon in the context of mantle-driven flow. Last, we investigate the importance of absolute plate motions for orogenic deformation.

Observations Lithospheric deformation The upper crust is the best-studied lithospheric layer in terms of deformation. Deformation is both aseismic and seismic, and the bulk rheology is mostly characterized by Coulomb behaviour. The deformation mechanisms for upper crustal deformation are cataclastic flow, pressure solution and dislocation creep. The translation and rotation of upper crustal blocks is well demonstrated by Global Positioning System (GPS) data. Internal strain of the upper crustal blocks is also recorded by GPS data, although the signal contains both transient elastic deformation of the crust and shallow mantle (i.e. seismic cycle) and permanent tectonic strain due to plate motions. Standard kinematic analysis of structural geological data allows one to constrain three-dimensional, finite deformation. Accurate average velocities (and strain rates) are obtainable with this technique, provided that the timing of deformation is well bracketed (e.g. Christensen et al 1989; Dunlap et al 1997). The lack of normal seismicity indicates that the lower crust (>15 km depth) is significantly different Theologically and/or compositionally from the upper crust. Geological observations on exposed lower crustal sections suggest a bulk felsic composition for the lower crust, consistent with seismic velocity (i.e. Vp/Vs) studies on in situ sections (Christensen & Mooney 1995). Layering in the lower crust is generally subhorizontal to shallowly dipping, observed both geologically and geophysically. In addition, large sections of the exposed lower crust contain nappes and sheath-fold structures, indicating large amounts of plastic strain (e.g. Goscombe 1992). These structures are important as they

imply that the lower crust is pervasively deformed. Mantle deformation can be deduced from shear-wave splitting datasets in a variety of orogenic zones. The result of seismic shear waves travelling through anisotropic media is shearwave splitting, similar to optical birefringence, in which the incoming shear wave is split in two quasi-orthogonal waves that travel at different speeds (see reviews by Silver 1996; Savage 1999). The time lag between the two waves is controlled by the intrinsic anisotropy of the medium, which depends on the intensity of the lattice preferred orientations, and by the thickness of the anisotropic layer. Typical splitting delays in neotectonic regions are ~l-2 s. Most of the observed shear-wave splitting for teleseismic travel paths is considered to arise in the upper mantle, extending down to the 410km seismic discontinuity. While crustal rocks contain large intrinsic anisotropies, the relative thinness and the variability of the deformation fabrics of the continental crust rule it out as a major source of shear-wave splitting. Barruol & Kern (1996) have shown that the crust typically contributes ^0.1-0.2 s of splitting, with 0.5s delays in extreme cases. Moreover, kilometrescale exposures of upper mantle rocks show a strong fabric, consistent with the interpretation of pervasive deformation in these settings (e.g. Christensen 2002). Mantle fabric Olivine LPO and shear-wave splitting. The observed shear-wave splitting in mobile belts is presumed to result primarily from the lattice preferred orientation (LPO) of olivine, the most abundant mineral in the upper mantle. Olivine displays anisotropic elastic properties (Abramson et al. 1997), with compressional seismic waves travelling 25% faster when parallel to the fast [100] axis relative to the slow [010] axis. Shear-wave propagation is also affected by this elastic anisotropy. The highest anisotropy (c. 18%) is recorded by waves propagating in a direction intermediate between the [100] and the [001] crystallographic axes, and fast split shear waves are polarized parallel to the [100] axis of olivine. The anisotropy of olivine is relatively low compared to that of some other silicate minerals, such as biotite. The large delay times recorded by shear-wave splitting are due to the thickness of the mantle sampled by the seismic waves, rather than to the inherent anisotropy of the material. Shear-wave splitting requires that the olivine crystals display coherent orientations (i.e. strong LPO) over large distances in the

MANTLE-DRIVEN DEFORMATION OF OROGENS

upper mantle (>50km). Therefore, dislocation creep must be active in at least the upper 250300 km of the mantle, since diffusion creep does not produce an LPO. This inference is consistent with experimental deformation, which suggests that dislocation creep occurs in olivine aggregates at upper mantle temperatures and pressures (e.g. Zhang & Karato 1995). Seismic anisotropy, such as shear-wave splitting, results from the deformational fabric of mantle rocks. Shear-wave splitting observations provide information about deformation in the mantle, allowing us to conduct structural geology analysis of the mantle in these settings. The relation between deformation and olivine LPO is critical to our interpretation of seismic anisotropy data in terms of upper mantle kinematics. Comparison between shape preferred orientation of olivine and spinels in a large number of naturally deformed mantle rocks indicates that the olivine [100] axis usually tends to concentrate at low angle to the lineation. The [010] axis either concentrates at high angles to the foliation or forms a girdle normal to the lineation (Ben Ismail & Mainprice 1998). This LPO suggests a dominant activation of the (010) [100] and (001)[100] slip systems in olivine for deformation under upper mantle conditions (Tommasi et al 2000). Analysis of olivine LPO evolution in high-temperature deformation experiments, as well as in polycrystalline plasticity models, provides some further constraints on this relation. Laboratory experiments. Deformation experiments of both olivine single crystals and polycrystalline aggregates under high-temperature, high-pressure conditions provide valuable constraints on: (1) how the activation of the various slip systems, and hence the LPO, is affected by physical parameters such as temperature, water and oxygen fugacity; and (2) how olivine LPO reflects the flow kinematics and/or finite strain. Flow experiments on olivine crystals, under predicted upper mantle pressure and temperature conditions, show that the dominant slip system is most likely (010) [100], followed by the (001)[100] system. Under high water or oxygen fugacity conditions, slip on the latter system tends to dominate (Bai et al. 1991; Mackwell et al. 1985). On the other hand, lowtemperature experiments show a dominant activation of glide on the [001] direction (Phakey et al. 1972). Analysis of olivine LPO in naturally deformed peridotites shows that upper mantle deformation is generally accommodated by slip on both (010)[100] and (001)[100] systems, with dominance of the former. Moreover, glide in the [001] direction is only significant during

43

low-temperature deformation associated with the emplacement of mantle slabs into the crust (Tommasi et al. 2000). In simple shear experiments under high-7 conditions, deformation fabrics yield seismic anisotropy with an LPO characterized by a concentration of olivine [100] axes parallel to the elongation direction (Fig. 1). The fabrics obtained in the various experiments deviate, however, with respect to the orientation of [010] axes and [001] axes. In plane-strain experiments, Zhang & Karato (1995) found point maxima for the crystallographic [010] axes and [001] axes of olivine approximately perpendicular to the shear plane and approximately perpendicular to the shear direction within the shear plane, respectively. In contrast, Bystriky et al. (2000) report girdles of crystallographic [010] axes and [001] axes of olivine for torsion experiments (Fig. 1). Both types of LPO are common in mantle xenoliths and mantle sections of ophiolites (Ben Ismail & Mainprice 1998). We tentatively suggest that the Zhang & Karato (1995) data represent slightly higher temperatures (1300°C v. 1200°C) than the experiments of Bystriky et al (2000). The LPO patterns are the same, only the development is faster for the Zhang & Karato experiments because of stronger dynamic recrystallization. This inference is broadly consistent with the results of Carter & Ave Lallemant (1970), in which point distributions of crystallographic axes and activation of (001)[100] occurred at the highest-temperature conditions. A major difference, however, between the Zhang & Karato (1995) and Bystriky et al. (2000) experiments is the type of strain. The experiments of Bystriky et al. (2000) are simple shear, whereas the experiments of Zhang & Karato (1995) contained a component of coaxial deformation parallel to the shear direction (reported in Zhang et al. 2000). This coaxial component, in addition to the possible formation of extensional shear bands, will lower the angle between the long axis of the finite strain ellipse and the shear plane. Therefore, in both experiments, the [100] axes are parallel to the long axis of the finite strain ellipsoid. Numerical experiments. The development of LPO in olivine aggregates has also been addressed using forward modelling. Several approaches have been utilized, including viscoplastic self-consistent (Wenk et al. 1991; Wenk & Tome 1999; Tommasi et al 1999), stress equilibrium (Chastel et al 1993) and kinematic constraint (Ribe & Yu 1991; Kaminski & Ribe 2001) models. These LPO predictions using these various approaches and their application to the

44

B. TIKOFFEJAZ,

Fig. 1. Olivine microstructures produced under different deformation boundary conditions. The experiments of Zhang & Karato (1995), potentially applicable to higher-temperature deformation, produce point maxima. The results of Bystriky et al. (2000) indicate a point maximum for the [100] axis and a girdle distribution for the [010] and [001] axes. The resultant shear-wave splitting, in either case, depends on the orientation of the LPO since SKS waves have a near-vertical incidence. In either case, vertical foliation and horizontal lineation produces strong shear-wave splitting.

prediction of seismic properties are reviewed in Tommasi et al. (2000). In general, classical polycrystalline plasticity models (in which deformation is accommodated by dislocation glide only) predict that the olivine [100] axes tend to parallel the maximum finite

elongation direction and that the [010] axis will tend to align normal to the foliation. In more complex three-dimensional flow types, the olivine [100] axis tends to align with the long axis of the finite strain ellipsoid (Tommasi et al. 1999). These results hold true specifically for

MANTLE-DRIVEN DEFORMATION OF OROGENS

transpressional and transtensional deformation. These deformations are unusual because a switch in lineation orientation (transpression) or foliation orientation (transtension) can occur for either: (1) different angles of oblique convergence/divergence; or (2) increasing strain at a single angle of oblique convergence/divergence (Tikoff & Greene 1997). Analysis of olivine LPO development in polycrystalline plasticity models also highlights that LPO intensity does not depend linearly on finite strain (Tommasi et al 2000). At shear strains of 1-2, clear LPOs are already formed. This result is in agreement with LPO evolution in simple-shear experiments, which shows a fast evolution up to shear strains of 1 or 2 and then maintenance of a steady-state fabric (Bystriky et al. 2000). Comparison between model predictions and LPO intensity in naturally deformed rocks also suggests that LPO only records part of the total strain accommodated by the rock. Mantle viscosity: the significance of seismic attenuation The degree of observed shear-wave splitting for mantle paths attains a global average of just over 1 s and is as high as 2.7 s (MarsonPidgeon & Savage 1997). These values imply that, for nominal values of intrinsic anisotropy derived from laboratory studies, coherent deformation fabrics in the upper mantle of the order of 100-200 km thick exist (Silver 1996). The vertical extent of this pervasive deformation has given rise to an important debate (e.g. Vinnik et al 1989a, b\ Silver & Chan 1991). Is this pervasive fabric contained in a highly viscous lithosphere, or does the shear splitting signal arise from a plastically flowing asthenosphere? Given normal geotherms, especially for tectonically active regions, 200 km of coherently deformed material may include an anisotropic asthenospheric source layer as well as the 'frozen' fabrics of the overlying lithosphere. If the anisotropic layer is strictly asthenospheric, then this ductile layer is potentially a zone of weak viscous coupling, or even decoupling, between the lithosphere and the deeper mesosphere. Seismic attenuation has the potential to resolve this issue. Seismic attenuation is due either to scattering of seismic energy along the wave path or to intrinsic attenuation of seismic energy caused by conversion of displacive energy into thermal energy (thermal dissipation). Normally, such loss is higher for shear waves than for compressional

45

waves. Therefore, a comparison of the frequency contents of S waves relative to P waves can be made to be diagnostic of attenuation in a quantitative way. This measure is referred to as '2' for quality factor. Q is a dimensionless number that is inversely related to attenuation. For Earth materials that are near their solidus, Q is normally in the 100-200 range. Q for rigid lithosphere of oceanic slabs is 800 or more (e.g. Okal & Talandier 1997). Consequently, Q measurements are useful indicators of rheology. Attenuation along paths restricted to oceanic lithosphere, excluding spreading ridges, is low (high Q). In cratonic regions, presumably underlain by thick, cold, high-velocity roots, seismic attenuation is also low. For these two areas, oceanic lithosphere and cratonic regions, the upper mantle is inferred to be relatively rigid (e.g. Lay & Wallace 1995). In contrast, for seismic travel paths that include oceanic asthenosphere or upper mantle beneath most actively deforming regions, attenuation is high (low Q). Seismologists typically interpret such regions to be near solidus temperatures and Theologically weak or even flowing. The presence of small percentages of partial melt (1-2%) is often discussed for these regions (e.g. Roth et al. 1999). The crustal contribution to attenuation is minimal, due to the short path lengths for nearvertical, teleseismic rays. The combination of shear-wave splitting measurements and seismic attenuation observations can constrain the ultimate source of mantle deformation. For example, one can discriminate between active mantle flow in a rapidly deforming layer and presence of a relict fabric in a lithospheric volume that is not deforming actively. In both cases, we might observe a strong, consistent shear-wave splitting pattern across an area. In the flowing asthenosphere case, however, attenuation is high (low Q). In contrast, attenuation will be low for the preserved old lithosphere fabric. If both strong lithospheric and asthenospheric fabric are present in an area, Q measurements may still discriminate between the two given the proper set of earthquake sources and seismic stations. The lateral extent of long delay-time shearwave splitting measurements in active tectonic regions suggests that the relatively narrow plate boundaries observed on the surface in continental regions correspond to very wide plate boundaries in the upper mantle (Russo et al. 1996; Klosko et al 1999). Assuming that this upper mantle material is flowing by crystalplastic processes, lateral displacements are potentially controlled by variations in lithospheric thickness. Convergent boundaries, by definition,

46

B. TIKOFFCTAL.

result from the juxtaposition of two different lithospheres. Consequently, mantle flow in these regions is affected by thickness variations in both lithospheres. Lithosphere/asthenosphere connections Cratons and asthenosphere Oceanic lithosphere contains a seismically detectable low-velocity zone inferred from dispersion of surface waves in ocean basins. Continental lithosphere, in contrast, lacks a low-velocity zone. Additionally, tomographic observations indicate that Precambrian cratonal regions are underlain by high-velocity regions extending to 300 km or more in some cases (e.g. van der Lee & Nolet 1997). These observations imply that: either (1) cratonic regions are coupled to large-scale mantle circulation (and hence there is no asthenospheric shear zone at their bases); or (2) any decoupling layer beneath cratons is much deeper than decoupling beneath ocean basins (~160 km) and is essentially different in seismic character. Compositional differentiation whereby a buoyant, strong residuum of mantle material forms the roots beneath ancient continental cores can account for the thicknesses, observed seismic velocity and apparent unsubductibility of the cratonic mantle lithosphere (e.g. Jordan 1975; Carlson 1995). These roots may also explain that asthenosphere does not develop at normal depths beneath these regions. If 'asthenospheric' shear zones do exist beneath cratons, the thermal regime, pressure regime and composition are significantly different from those of oceanic asthenosphere. Therefore, significantly different rheology and strain rates are expected. Given the possible viscosity increases, the effects of these differences on rheology are difficult to constrain quantitatively. Insight into the deformation regime beneath cratons can be garnered from mantle xenoliths brought to the surface at kimberlite pipes. Kennedy et al (2002) note that mantle peridotites that demonstrably come from the deep subcratonic mantle are divisible into two groups: (1) coarse-grained protogranular peridotites, and (2) fine-grained porphyroclastic peridotites. The latter are pervasively deformed at higher strain rates than the former and are only found at lower lithospheric levels (~ 150-200 km) as determined from thermobarometry. Kennedy et al (2002) infer that deformation recorded by the porphyroclastic peridotites is more recent than the last-known deformation (Phanerozoic) visible at the Earth's surface. What younger deformation can these fabrics possibly represent?

These authors argue that the porphyroclastic peridotites found in the kimberlite xenoliths are samples of the lithosphere-asthenosphere decoupling shear zone beneath cratons, thus explaining their high strain rate deformation (Kennedy et al. 2002). If this is correct, then shear at the base of cratons occurs deeper and without the associated seismic low-velocity zone found in oceanic asthenosphere. There are several potential problems with this interpretation, including the lack of independently confirmed textural ages of the fabrics and the cause of initial formation. Finally, we note that the 'high' strain rate inferred from these xenoliths may be only relatively high, in comparison to those expected in oceanic asthenosphere. Thus, the very low seismic attenuation typically observed in cratonic regions, and the absence of a seismically defined low-velocity zone, could both be consistent with a deep, lower strain rate acting within a decoupling zone beneath cratons. Moreover, subcratonic mantle shear zones have sufficient time to acquire such a fabric. In contrast, the fabrics developed in oceanic asthenosphere are limited by the temporal existence of unsubducted young oceanic lithosphere (< 170 Ma). Lithospheric v. asthenospheric fabrics Oceanic settings. The relation between lithospheric and asthenospheric upper mantle 'deformation' fabrics is clearest in the Pacific Ocean basin. Pn waves (Moho head waves, which are a regional seismic phase that travels along the Moho for most of its propagation) travelling just below the oceanic crust exhibit strong azimuthal anisotropy (e.g. Hess 1964; Raitt et al. 1969). The fast trends for these phases align with the mineralogical fabric imparted at spreading ridges when magma freezes while undergoing shear at the base of forming lithosphere (Fig. 2). This shear flow aligns upper mantle olivine [100] axes parallel to the spreading direction. In the older portions of the Pacific basin, this alignment results in fast Pn trends parallel to fracture zones. In contrast, a small but growing dataset of teleseismic shear-wave splitting indicates that this anisotropy is not detected. Instead, a strong asthenospheric signal yields fast shear polarization directions parallel to the current absolute plate motion (hotspots assumed fixed) of the Pacific plate (Fig. 2; Russo & Okal 1998; Wolfe & Silver 1998; Klosko et al 1999). Surface wave analyses of the ocean basins show that upper mantle azimuthal anisotropy is generally closely related to lithospheric absolute plate motions, particularly in the Pacific basin (Nishimura & Forsyth 1988, 1989; Montagner &

MANTLE-DRIVEN DEFORMATION OF OROGENS

Fig. 2. Shear-wave splitting in oceanic material, in map view. Lithospheric fabric, frozen in from asthenospheric flow at the spreading centre, is parallel to the fracture traces. Asthenospheric fabrics are parallel to the current direction of plate motion in a fixed mantle reference frame. These two diverge if the motion of the plate has changed, such as for oceanic material older than 43 Ma in the Pacific region.

Tanimoto 1990, 1991). Pn anisotropy and shearwave splitting are generally not parallel in the Pacific basin, since the fracture zones in lithosphere older than around 43 Ma do not parallel today's absolute plate motion direction. These data have been interpreted to signal the operation of simple asthenospheric flow beneath the Pacific plate (Nishimura & Forsyth 1988,1989; Montagner & Tanimoto 1990, 1991; Russo & Okal 1998; Wolfe & Silver 1998; Klosko et al 1999; Vinnik et al 1989a, b). As the plate moves rapidly towards western Pacific subduction zones, shear is imposed by the underlying asthenosphere. This shear transfer is presumed to occur in the low-velocity zone, producing effective decoupling or weak viscous coupling between the plate and the upper mantle below the asthenosphere. Fabric in the lithospheric mantle is not properly a lithospheric deformation fabric in the sense we have been using elsewhere in this chapter. It records a 'frozen' asthenospheric flow that formed at a spreading centre, and not lithospheric deformation. This is distinct from the current fabric in the asthenosphere, which formed since 43 Ma when the Pacific plate's motion changed. Thus, asthenospheric fabric beneath oceanic lithosphere is potentially erased in an oceanic setting. This 'erasing' may not occur in continental lithosphere-asthenosphere fabrics (Vauchez & Garrido 2001).

47

Continental settings. The more complicated horizontal layering of the continental lithosphere-asthenosphere system produces a different response to tectonic loading. Lateral loading has the three end-member cases of contractional, extensional and wrench loading. Each type of loading results in a different mechanical response: thrusting, viscous shortening, thickening and buckling (contraction); normal faulting, crustal boudinage and extensional necking (extension); and strike-slip faulting and ductile shear flow (wrench). However, the observational basis for comparing lithospheric and asthenospheric fabrics in continental lithosphere is lacking for two reasons. First, discrimination between lithosphere and asthenosphere in continental settings requires more sophisticated seismic measurements than standard travel time inversion. Second, the heterogeneity of the continental lithosphere makes resolution of the lithosphere-asthenosphere difficult. For example, tomography might reveal what appears to be a cold, stiff lithosphere overlying a hotter asthenosphere, but a compositional difference could just as well explain the two velocity anomalies. Attenuation measurements for phases following specific paths, such as Pn or Sn, can afford a basis for discrimination between layers that are Theologically strong or weak. However, interpretations are tenuous without earthquakes or well-placed stations. Pn (Moho head) waves appear not to propagate in geologically complex regions, although the exact reasons are unknown. This leads to an unsatisfying guess about the actual rheology at depth. Regions that are tectonically active are least likely to have thick lithosphere (e.g. Plomerova et al. 2002), and thus measurements of core phase shear-wave splitting (e.g. SKS) are likely to sample asthenospheric anisotropy only. Even if the lithospheric signal is clearly discriminated from the asthenospheric signal, the paths of shear waves most commonly used for splitting analyses cross both, yielding an inherent uncertainty in the source(s) of any splitting observed at the surface. In a few continental regions, systematic variability in splitting parameters was detected and interpreted as a function of the arrival azimuth of the shear wave at the station (Savage & Silver 1993). Such variability is most often ascribed to the presence of two layers of different anisotropy, each of which splits the arriving shear waves. Interpretations of such data have so far been restricted to assuming that the topmost layer is the lithosphere (or crust) and the lower layer is the asthenosphere (or possibly the mantle lithosphere) (Savage & Silver 1993). The inherent uncertainty arises from the mismatch between

48

B. TIKOFFE7AL.

two assumed layers of anisotropy, although there are four potentially different anisotropic layers to consider (upper crust, lower crust, mantle lithosphere and asthenosphere). As the crustal component contributes only 0.1 -0.2 s of splitting because of its limited extent (Barruol & Kern 1996), the mantle layers should dominate the splitting. Where two-layer anisotropy is detectable, different fabrics must be stacked vertically, which requires strong vertical deformation gradients. An advanced type of study involves simultaneous determinations of Pn velocity and anisotropy (Hearn 1996, 1999; Mele et al 1998; Calvert et al. 2000). Pn can provide a useful constraint on the question of lithosphere asthenosphere fabric differences. This phase preferentially samples the uppermost mantle, which is probably lithosphere even in active tectonic environments with thin lithospheres. Estimates of azimuthal anisotropy from Pn can be directly compared to core phase shear-wave splitting measurements, which generally sample the entire upper mantle, including the lithosphereasthenosphere package. Furthermore, seismic investigations require a trade-off between Pn anisotropy and lateral velocity (Mochizuki et al. 1997), compromising fine resolution. However, careful studies show interpretable results (Hearn 1996). Good Pn anisotropy is comparable to shear-wave splitting results, and can potentially determine whether the splitting is dominated by the uppermost mantle portion of the phases' travel paths or deeper portions thereof. If the Pn and splitting anisotropies are similar, the vertical distribution of anisotropy must be fairly constant. Moreover, if mantle lithosphere and asthenosphere are both present, the fabrics of each must be similar. Alternatively, if the anisotropies differ, it is likely that the lithospheric fabrics differ from asthenospheric fabrics. In many regions, comparison of Pn- and splitting-derived anisotropies indicates that the uppermost mantle fabrics are very similar to the integrated fabrics delineated by shear-wave splitting. The tectonic environments in which this is true include the region around the San Andreas fault and the western portion of the Great Basin (Hearn 1996), the Cascades region of western Washington state (compare Hearn 1996 and Bostock & Cassidy 1995), the central Appenines (Margheriti et al. 1996; Amato et al. 1991 \ Mele et al. 1998; Hearn 1999) and the Aegean region (compare Hearn 1999 and Hatzfeld et al. 2000). Thus, regions dominated by contraction, extension and wrenching all include large regions where uppermost lithosphere and deeper lithosphere-asthenosphere packages

have similar fabrics in the Theologically layered continental stack. Note that the correspondence between Pn and splitting anisotropy is not always close, even in the small portion of the continents that have been carefully studied. Thus, there is the possibility for differential vertical deformation between the lithosphere and asthenosphere. Regions that document disagreement between the two sets of observations include the Sierra Nevada region and central valley of California, the Idaho Batholith-Snake River Plain region, the Rio Grande rift area (Hearn 1996), the Tyrrhenian Sea (Mele et al. 1998; Hearn 1999) and the Rhine Graben (Vinnik et al. 1994; Enderle et al. 1996; Plomerova et al. 1998; Hearn 1999).

Coincidence of crustal deformation and mantle fabric Ancient orogens In a variety of settings, upper crustal and mantle deformation are remarkably similar. In all cases, mantle deformation is constrained by shear-wave splitting data. The deformation of the upper crust is recorded by a variety of features. In ancient orogens, the record of upper crustal deformation is typically that of the orientation of structures. The Superior Province of North America and the Kaapvaal craton of South Africa both show shear-wave splitting parallel to the dominant Archaean-age structure of the region (e.g. Silver 1996). Likewise, parallelism between Late Proterozoic collisional fabrics in the mid to lower crust and SKS splitting data are also recorded in the Ribeira-Aracuai belt of SE Brazil (Vauchez et al. 1994; James & Assumpcao 1996). Palaeozoic orogenies show similar patterns. Barruol et al. (1997) show that the shearwave splitting parallels the structural trend of the Appalachian-Caledonian-Hercynian belt. Consequently, for this and all the above cases, the current shear-wave splitting data represent a 'frozen' signal in the mantle lithosphere from orogenesis. Vauchez et al. (1997) further suggest that the reactivation of orogenic belts in continental break-up is a result of this 'frozen' anisotropy in the mantle lithosphere. Poly crystalline plasticity models show that a mechanical anisotropy, due to olivine LPO frozen in the lithospheric mantle since the last orogenic episode, may induce a directional strain softening in the lithospheric mantle and control the localization of continental break-up (Tommasi & Vauchez 2001).

MANTLE-DRIVEN DEFORMATION OF OROGENS

Neotectonic orogens Neotectonic zones allow more quantitative conclusions to be drawn on the relation between the development of mantle fabric and crustal deformation. A wide range of neotectonic areas have been studied, ranging from transcurrent domains (Trinidad) to convergent ones (Himalayas). The San Andreas system and New Zealand represent intermediate situations, characterized by increasing amounts of convergent motion in obliquely convergent settings (Fig. 3). Despite the wide range of tectonic settings, there are clear similarities in all of the above examples. Regardless of the angle of relative plate motion, the fast polarization direction of the shear-wave splitting is often at a small angle to the plate boundary. This is particularly clear in the obliquely convergent boundary in New Zealand (Fig. 3; Klosko et al 1999; Molnar et al. 1999). The stations displaying the highest delay times show a fast shear wave polarized at ~10-15 degrees from the trace of the Alpine fault. As one moves away from the plate boundary, the magnitude of the time delay decreases and the directions are oriented at higher angles to the Alpine fault. This observation indicates that some rotation in the orientation of LPO occurs as a result of finite strain gradients. Patterns of shear-wave splitting show ubiquitous rotation of splitting directions near strike-slip faults (San Andreas fault, California; El Pilar fault, Venezuela) in obliquely convergent systems. This observation suggests that zones of strain localization in the upper crust correspond to zones of strain localization in the upper mantle (e.g. Molnar 1992). In the Himalayas, fast shear-wave polarizations are only parallel to the trend of the belt close to the major strike-slip faults (Fig. 4; Lave et al 1996). The general orientation of shear-wave splitting subparallel to the plate boundaries suggests: (1) rotation of the finite strain ellipse to an orientation parallel to the boundary; or (2) lateral flow along the margin (plane transpression of Tommasi et al 1999). Interpretation of the data Holt (2000), using geodetic data from Tibet, calculated maximum shear strain rate planes. One of these planes is typically subparallel to the observed shear-wave splitting. Three different approaches were tried in South Island, New Zealand. Moore et al (2002) utilized the orientation of the present velocity field and assumed that the deformation had accumulated for ~6.5 Ma with this velocity field. The resultant deformation fits well with the

49

observed shear-wave splitting. Little et al (2002) calculate surface deformation using deformed, crustal-scale markers, and also find a very good correlation to inferred mantle deformation. In general, the surface deformation and shearwave splitting data are consistent in extent and magnitude. It is not yet clear how to interpret this coincidence. There are at least two possibilities: (1) the mantle and the crust are coupled (McKenzie & Jackson 1983; Molnar 1992; Teyssier & Tikoff 1998); or (2) similar boundary conditions apply to the upper crust and upper mantle (Holt 2000). A third possibility, simple asthenospheric flow below the continents (e.g. Silver 1996), does not apply to orogenic zones. An exception to the last possibility is the proposed flow of asthenosphere under the Basin and Range of the western United States (e.g. Silver & Holt 2002). The possibility that both lithospheric layers respond independently to side-driven boundary conditions (e.g. Holt 2000) is unlikely for a variety of reasons. First, the upper crust, as an individual lithospheric layer, is unlikely to transmit the stresses from the plate boundary to the interior of Asia. This is particularly true for the Himalayas given the existence of multiple lithospheric-scale suture zones bounding the collisional terranes, which would act as zones of strain localization. Second, Giorgis et al (2004) document the coincidence in upper crustal deformation and mantle deformation associated with vertical-axis rotations in dominantly transcurrent environments. They document that bottomdriven models, compared to side-driven models, better describe the amount of vertical-axis rotation relative to the orientation of shearwave splitting. The most significant, however, is the nature of mechanical instabilities within independently deforming layers. The thin and laterally extensive geometry of the lithospheric plates allows for only a few styles of deformation in response to a horizontal end load, such as buckling or boudinage (e.g. Burg et al 1994 for the Himalayas). These types of mechanical instabilities cause localization of strain, which leads to discontinuous deformation. However, geodetic surveys of most orogenic zones indicate that upper crustal flow is broadly continuous (e.g. Beavan & Haines 2001 for the South Island of New Zealand). Consequently, it is unlikely that the upper crust deforms independently of the rest of the lithosphere. An equally viable alternative is that the layers are in mechanical communication. This requires that the lower crust, and other presumed weak lithospheric horizons in the lithosphere, act to partially couple the upper crust and lithospheric

50

B. TIKOFFCTAL.

Fig. 3. The relation of absolute plate motion (APM), relative plate motion (RPM) and shear-wave splitting (dark lines) for four neotectonic areas. See text for details.

MANTLE-DRIVEN DEFORMATION OF OROGENS

51

Fig. 4. Cartoon of lithospheric deformation in Tibet. Dark lines indicate the orientation and magnitude (length) of shear-wave splitting measurements. The short, wavy lines indicate zones of ductile shearing. Crustal deformation is characterized by normal faulting and strike-slip faulting, in addition to contractional structures. Mantle deformation, inferred from shear-wave splitting, indicates components of sinistral simple shear, concentrated near strike-slip faults, and coaxial extrusion (left side of diagram - shown with kinematic axes a, b and c). The coincidence of crustal deformation, inferred from geodetic motion in addition to geological structures, with the mantle deformation indicates that the lithospheric layers are coupled.

mantle layers. We explore this option in the next section (Clutch Tectonics). A major limitation: strain history Little et al (2002) suggest that the geological history is potentially recorded in seismic anisotropy. Using two markers that were offset by the Alpine fault in South Island of New Zealand, the Junction magnetic anomaly and the Moonlight fold belt, the finite strain was modelled for upper crustal deformation assuming a transpressional deformation model. The calculated upper crustal deformation is in very good agreement with the shear-wave splitting. While this approach has the disadvantage of the assumption of originally linear markers, it has the advantage of addressing the geological history. Little et al. (2002) demonstrated that the upper crustal deformation was probably not constant with time, consistent with plate motion reconstructions. Therefore, it is potentially incorrect to determine the currently active deformation (e.g. geodetic data) and then integrate the data over some time to calculate the

shear-wave splitting. Because shear-wave splitting results from a fabric, it is critical to incorporate the geological history of the plate boundary. Clutch tectonics We hypothesize that deformation in the crust and mantle are similar because these lithospheric layers are mechanically connected. Therefore, either displacements initiate in the upper crust and move downward (top-driven system) or displacements initiate in the mantle and move upward (bottom-driven) (Fig. 5). The zone that connects the top to the bottom is called a clutch zone, by analogy to the clutch of an automobile (Tikoff et al. 2002). A clutch allows mechanical continuity between the independently operating wheels and engine. Clutch zones, when present, have a different shear sense for top-driven or bottom-driven systems (Fig. 5c v. 5d). While the concept of top-driven systems was investigated for extensional deformation (e.g. Axen et al. 1998), bottom-driven systems must be the general rule in orogenic zones if tectonic

52

B. TIKOFFCTAL.

Fig. 5. Vertical, cross-sectional views of the relation between crustal and mantle deformation. The indicator represents an imaginary, initially vertical line that changes orientation with deformation (a loose line). The crust moves the same amount in all cases, but the relation between the upper crust and the mantle changes. Detachments (b) cause no fabric and are not consistent with the coincidence of crust and mantle deformation. Top-driven systems (c) produce the opposite sense of shear from bottom-driven systems (d,e). The lower crust can be involved (e), or not (d), in a bottom-driven system. Bottom-driven systems are expected to be the norm in orogenic belts. Modified from Tikoff et al (2002).

movements ultimately result from deep-mantle processes. Bottom-driven systems do not require particular wavelengths of deformation, as the deformation does not proceed by amplification of instabilities. Rather, deformation is semicontinuous, an observation which is matched in many orogenic zones (e.g. Beavan & Raines 2001). Top-driven systems The Pacific Ocean lithosphere-asthenosphere (Fig. 2) and other ocean basins are potentially a

good example of a top-driven tectonic system. In this case, the movement of the oceanic lithosphere may be governed by boundary forces affecting the edges of the plates (ridge push or slab pull), which affects the flow of the asthenosphere. Available seismic anisotropy data imply a velocity gradient between the plate and the mesosphere, and it is assumed that the higherviscosity mesosphere moves more slowly. Regardless, as the lithosphere is pulled toward a subduction zone, the gradual variation in viscosity that causes the rheological layering of the lithosphere and asthenosphere will lead to a

MANTLE-DRIVEN DEFORMATION OF OROGENS

wide zone of decoupling. Note that this topdriven system does not deform the oceanic crust away from the ridge. This rigidity probably results from the homogeneous nature of oceanic crust compared to continental crust. Bottom-driven systems The coincidence of upper crustal and upper mantle deformation in continental settings suggests that the processes in these two layers are coupled. Movement must ultimately result from mantle convection driven by thermal dissipation, and thus the likely outcome of coupled mantle and crustal deformation is that orogenic systems are bottom-driven (e.g. Molnar 1992). The formation of fabrics in the lower crust is also consistent with a notion of partial attachment (Tikoff et al 2002). Even complete attachments result in the formation of different fabric, as the lithospheric layers have differing rheologies. Complete decollements occur in Nature and require zones of mechanical decoupling, which are generally discrete (e.g. salt horizons in a thrust system). The ubiquitous occurrence of originally subhorizontal, lower crustal layering and structural indications of high strain (intrafolial folds, sheath folds) suggests mechanical communication between the upper crust and upper mantle. Consequently, not only is coupled surficial (upper crustal) and mantle deformation explained by clutch tectonics, but so is the formation of lower crustal layering (Tikoff et al. 2002). The major question of what causes the flow patterns in orogenic zones remains. If one of the lithospheres involved in plate interaction is relatively strong, such as an oceanic or strong cratonal region, deformation will preferentially occur in the weaker region, which will eventually become a deformation zone. The effect of lithospheric roots (from cratons, magmatic arcs, etc.) is therefore to channel asthenospheric mantle flow. The three-dimensional flow in the mantle within these deformation zones is thus controlled by relative plate movement. This movement, in turn, causes pressure gradients in the mantle and lithospheric thickness variations. Finally, part of the simple shear component parallel to the plate margin is probably an intrinsic part of the global (convective) mantle flow. It is clearly obtained in convection models, provided that the plate boundaries are characterized by a very efficient strain softening.

Implications for tectonics Based on surficial crustal movements, interpretation of lithospheric and asthenospheric mantle

53

fabric, and inferences of lower crustal deformation, we propose three-dimensional models for obliquely convergent, transcurrent and obliquely divergent boundaries. These are shown in block diagrams in Figure 6. We first discuss the strain patterns (e.g. Fossen & Tikoff 1993; Fossen et al. 1994) and the resultant LPO (e.g. Tommasi et al. 1999) that we would expect if the deformation were homogeneous. We then compare these predictions to the observations of crustal deformation and shear-wave splitting. We wish to emphasize that these models are not strictly based on lithospheric strength profiles. Although these strength profiles are the basis for many tectonic models, they are extrapolations of laboratory data that may or may not be useful guides to rock behaviour at geological conditions. While strength profiles are widely utilized, our approach is to relax these criteria to put together an alternative, empirically based model. Additionally, we do not explicitly incorporate other models, such as lower crustal channel flow (e.g. Roy den et al. 1997), which are derived principally from the strength profiles. Transcurrent boundaries (Fig. 6a) Wrench deformation makes a very simple prediction about the formation of fabrics. The material is essentially deformed within a simple shear zone, with the shear zone oriented vertically and parallel to the plate margin. In this setting, plane strain fabrics are expected to develop. Foliation is vertical and lineation is horizontal (foliation is defined by the maximum and intermediate axes of finite strain; lineation by the maximum axis of finite strain). Although both fabrics are initially oriented at a 45° to the shear plane, they rotate into parallelism with increasing finite strain (e.g. Lin & Williams 1992). The fabric in the mantle that corresponds to this style of deformation depends on the exact process of formation of the LPO (Fig. 1; e.g. point distributions v. girdle patterns). Regardless of the model chosen, strong shear-wave splitting from this LPO is predicted for vertical foliation and horizontal lineation (Fig. 1). The predictions of Tommasi et al. (1999) for wrench deformation are shown in Figure 7. Transcurrent boundaries have a very straightforward kinematic relation with respect to lithospheric coupling. Major strike-slip faults within the continental lithosphere cut the entire crust and continue into the mantle lithosphere (Stern & McBride 1997; Hole et al 1998). Fast polarization directions are typically subparallel or at a low angle to major active faults, including the San Andreas fault (California; Ozalaybey &

54

B. TIKOFF£TAL.

Fig. 6. Block diagrams of idealized deformation of the lithosphere. Wrench deformation (a) occurs in transcurrent plate boundaries and results in plane strain fabrics. Oblique divergence results in wrench-dominated transtension (b) or pure shear-dominated transtension (c), both of which result in constrictional fabrics. Pure shear-dominated transtension causes horizontal foliations. Pure shear-dominated transpression (d) and wrench-dominated transpression (e) result from oblique plate convergence. Both are associated with flattening fabrics. Pure shear-dominated transpression should result in vertical lineations, which cause low amounts of shear-wave splitting in the absence of lateral extrusion. Deformation that is dominantly divergent or convergent (e.g. pure shear-dominated) will tend towards plane strain fabrics.

MANTLE-DRIVEN DEFORMATION OF OROGENS

Fig. 7. Deformation types and the orientation of crystallographic axes of olivine, from predictions of a polycrystalline plasticity model valid for high-temperature (>900 °C) deformation in the mantle. The deformation types are shown as blocks on the left side of the diagram. Lower-hemisphere, equal-area stereonets are shown for the olivine crystallographic axes for 1000 grains contoured at 1% intervals. The black box shows the maximum LPO in each stereonet. Two predictions of LPO are shown for transpression, distinguishing wrench-dominated (top - long axis of finite strain ellipsoid is horizontal) and pure shear-dominated (bottom - long axis of finite strain ellipsoid is vertical). In general, olivine LPO is subparallel to the finite strain axes. Modified from Tommasi et al. (1999).

55

56

B. TIKOFF £7AL.

Savage 1995), El Pilar fault (Venezuela; Russo et al 1996), Kunlun fault (Tibet; McNamara et al. 1994) and Altyn Tagh (Tibet; Herquel et al 1995). Note that, in the case of the San Andreas fault, only the topmost of two layers of anisotropic material is oriented at a low angle to the San Andreas fault (Ozalaybey & Savage 1995). For the San Andreas and New Zealand cases, the mantle fabric appears to rotate towards parallelism with the boundary as one approaches the fault. This observation suggests that the deformation gradient is highest under the fault, but that deformation in the mantle occurs broadly under these plate boundaries. The same is true for the crustal deformation, as all of the transcurrent motions of these boundaries are not accommodated on the major, plateboundary faults. Diffuse crustal deformation accommodating transcurrent motion was proposed for the San Andreas system by Jamison (1991) and Teyssier & Tikoff (1998), and is observed geodetically and structurally for the South Island of New Zealand (e.g. Beavan & Raines 2001; Little et al 2002). Similar patterns are inferred from major, ancient strike-slip faults, such as the Great Glen fault (Scotland; Helffrich 1995) and strike-slip zones in the Appalachians (Barruol et al 1997). As addressed earlier, this fabric survives, and potentially even controls, subsequent continental break-up (Vauchez et al 1991 \ Tommasi & Vauchez 2001). Divergence (Fig. 6b, c) Oblique divergence will typically result in transtensional deformation, provided that a straight margin exists with a consistent oblique divergence movement direction. Transtension deformations typically lead to constrictional fabrics, although the exact nature of the fabric depends on the angle of oblique divergence and the amount of deformation (Fossen & Tikoff 1993; Fossen et al. 1994). Wrench-dominated transtension occurs for angles of oblique divergence less than 20°. Except settings that are within a few degrees from pure transcurrent motion, wrench-dominated transtension will result in very constrictional deformation. For pure sheardominated transtension and higher angles of oblique divergence, the deformation will start to approach plane strain controlled by the divergent component of motion. The lineation is parallel to the oblique movement direction and not the shear zone boundary. Foliation is vertical for low finite strain and low angles of oblique divergence. For any oblique divergence direction greater than 20°, the foliation is always horizontal (e.g. Fossen et al 1994). The LPO of olivine

tracks these directions well (Fig. 7), as determined by a polycrystalline plasticity model valid for high-temperature (>900°C) mantle deformation (Tommasi et al 1999). In general, less shear-wave splitting is expected for obliquely divergent settings, because of the horizontal foliation (Fig. 1). Divergence in continental lithosphere results in extensional or transtensional deformation, depending on the angle of oblique divergence. Mantle deformation shows characteristic fabrics with the fast split wave polarized oblique to the regional structural trend. The Rio Grande (Sandvol et al 1992) and East Africa (Kenya; Gao et al 1997) rifts both display this pattern. Lithospheric thinning may follow the structural expression of the rift, as demonstrated in Kenya (Achauer et al 1994). Upper crustal deformation responds to this deformation by undergoing a combination of strike-slip and normal faulting (e.g. Doser & Yarwood 1991). The similarity of upper crustal deformation patterns is closely consistent with experimental models of bottomcontrolled deformation (Withjack & Jamison 1986). Polycrystalline plasticity models predict that olivine LPO developed in response to a transtensional deformation will be oblique to the rift trend, giving rise to a seismic anisotropy signal characterized by fast split shear wave polarized in a direction oblique to the rift elongation (Tommasi et al 1999). This is consistent with the orientation of the shear-wave splitting data observed in the continental rifts. In our model (Fig. 6b, c), up welling and outward flowing asthenosphere provides the basal drag to cause the formation of fabrics in the lithospheric mantle. As regions experiencing oblique divergence typically have high heat flow, such as the Basin and Range region of the western United States, it is assumed that only the uppermost part of the mantle forms an LPO. This occurs because high temperatures tend to favour diffusional creep, which does not produce an LPO. We have not included passive rifting and/or orogenic collapse in the above model, despite the fact that these processes undoubtedly exist. Thickening of the crust and lithosphere inevitably introduces thermo-mechanical effects, such as thermal relaxation and time-dependent heating. These effects would act to decrease the coupling between the crust and the upper mantle (Vanderhaeghe & Teyssier 2001). These extensional systems may undergo top-driven collapse (e.g. Axen et al 1998), bottom-driven collapse, or some combination, with the variation occurring on both a spatial and temporal basis (e.g. Rey et al 2002; Tikoff et al 2002). The

MANTLE-DRIVEN DEFORMATION OF OROGENS

connection between these processes and mantle flow requires more study. Convergence (Fig. 6d, e) Convergence plate boundaries include oceanic oceanic, continental-oceanic, or continentalcontinental settings. We will concentrate on oceanic-continental and continental-continental settings. Oblique convergence results in transpressional deformation. Transpression causes flattening fabrics, which approach plane strain (pure shear) at very high angles of convergence. Foliation is always vertical, although lineation can be horizontal or vertical. Horizontal lineations, which result in strong shear-wave splitting, occur for an oblique convergence direction less than 20° and for low strains (e.g. Fossen & Tikoff 1993; Fossen et al 1994). The lineation starts at less than 45° to the shear zone boundary (plate boundary) and rotates into parallelism with the shear zone boundary at higher finite strain. Poly crystalline plasticity models (Tommasi et al. 1999) show good agreement between olivine LPO and finite strain axes (Fig. 7). Consequently, strong shear-wave splitting is predicted for the case of horizontal lineations. For the pure shear end-member, however, the vertical lineations provide very low to no shearwave splitting. Convergent settings show the largest difference between inferred mantle fabric from seismology and crustal deformation. In most orogenic zones, reverse faulting and associated folding show large amounts of crustal contraction perpendicular to the trend of the belt. In contrast, shear-wave splitting is generally perpendicular to the maximum shortening direction, for both neotectonic (Tien Shan, Wolfe & Vernon 1998) and ancient examples (Hercynian, Bormann et al. 1993; Ribeira belt, Heinz et al. 2003). Note, however, that some measurements are parallel to the maximum shortening direction in the above cases. From these data, mantle deformation is inferred to extend material perpendicular to the contraction direction and parallel to the trend of the belt. As indicated by Tommasi et al. (1999), this pattern cannot result from a pure shear deformation in the mantle in the plane of the shortening direction. Pure shear would create a vertical long axis which is incompatible with the large (>1 s) shear-wave splitting observed in these areas. Ways of interpreting the data include: (1) simple shear deformation; (2) pure shear deformation acting in a horizontal plane (lateral extrusion); or (3) a combination of the

57

two (Fig. 4). Geodetic and structural studies are not consistent with large amounts of strike-slip movement in many foreland fold and thrust belts. In the Himalayas, shear-wave splitting increases as one approaches the major strike-slip faults, including the Kunlun and Altyn Tagh (Fig. 4). The other possibility is coaxial deformation, lateral extrusion, which elongates material parallel to the strike of the fold and thrust belt. (We use 'lateral extrusion' to indicate a pure shear that both elongates and shortens material in a horizontal plane, although other kinematic possibilities for lateral extrusion are possible.) Lateral extrusion would explain the nearly exact perpendicular relation between shear-wave splitting and crustal contraction. Lateral extrusion requires some level of partial detachment (i.e. clutch tectonics). The shearwave splitting data indicate that the upper mantle is undergoing contractional flow below the thrust belt, in a similar direction to the crustal deformation. However, because of the different boundary conditions, the direction of extension is different for the crust and mantle. Because of the free surface of the Earth, the extension direction is upward and crustal deformation is characterized as reverse faulting. In the mantle, material cannot move downward easily, as it must displace other material. Therefore, the mantle moves laterally. Alternatively, both simple shear and a component of lateral extrusion act together (plane transpression). If extrusion tectonics is applicable to the Himalayan collision (e.g. Tapponnier et al. 2002), it is worth considering the role of mantle flow, in a bottom-driven system, and its effect on crustal deformation. The crustal deformation is only partially coupled from this extrusion, and accommodates more contractional deformation. A normal fault system occurs over most of the Tibetan plateau, which strikes north-south (parallel to the contraction direction) (Armijo et al. 1986; Yin et al. 1999). These normal faults are potentially recording a component of the extension deformation imposed by the extruding mantle flow in addition to lower crustal flow (Fig. 4). However, it is useful to remember that strain is a result of displacement gradient. This requires that the displacement from mantle extrusion increases from west to east (or, possibly, although less likely, from east to west) across Tibet. While the above discussion does not prove that the Himalayas are a bottom-driven system, it does indicate that a bottom-driven system is kinematically viable.

58

B. TIKOFFCTAL.

Strike-slip partitioning and homogeneous mantle deformation Strike-slip faults, which parallel the trend of the orogen, occur in most mountain belts. Although described originally for obliquely convergent arc settings (e.g. Fitch 1972), strike-slip faulting also occurs in continental collisions, such as the Himalayas, with dominantly convergent motion (e.g. strike-slip partitioning; Teyssier er <2/. 1995). If the upper crust is coupled to the mantle, the zones of strike-slip motion in the upper crust must represent zones of strong displacement gradients in the mantle. The physical experiments of Richard & Cobbold (1990) are a good analogy for this deformation. In these experiments, a layer of sand overlays a viscous silicone layer and the deformation was controlled by the bottom of the experiment. The sand developed a strike-slip fault immediately above maximum gradient of displacement in the underlying silicone. The sand on either side of the fault was deformed, often with reverse faults, but which accommodated a component of wrenching. If the upper mantle partitioned deformation at a scale smaller than a seismic wave, one would expect shear zones parallel to the plate margin accommodating the transcurrent motion and lithons between the shear zones accommodating the contraction (Fig. 8). As the shear waves would preferentially travel through the larger lithons, this would cause fabric parallel to the major faults. The absence of this pattern (e.g. South Island of New Zealand) indicates that the mantle does not partition deformation into discrete high-strain zones. Rather, the correspondence of crustal deformation and shear-wave splitting requires that the mantle deforms in a relatively homogeneous manner. Strike-slip partitioning, and most other crustal features in erogenic zones, may thus depend more on the mantle flow than local strength heterogeneities in the crust. The kinematic connection between movement in the mantle and crust may occur despite the major difference in the style of deformation: the upper crust deforms by faulting while the mantle deforms by crystalplastic deformation. Although plate-margin-parallel, strike-slip faults are locally necessary to accommodate mantle flow, and they are not capable of accommodating contraction or extension. Consequently, large regions of contractional or extensional deformation, exhibiting reverse or normal faults, must accommodate the remaining part of the imposed mantle displacement. Different types of faults (e.g. strike-slip and reverse) must coexist to accommodate brittle deformation driven by movement of a

ductile substrate. Consequently, the concept of a single regional stress direction that typifies an orogen is not compatible with a bottom-driven system. The simple relation between stress and flow may occur exclusively in the mantle, while the crust deforms passively in response to an imposed basally driven displacement field. To use the vernacular, the interesting physics may all be in the mantle while the crust rides along.

Absolute plate motion (APM) v. relative plate motion (RPM) A potentially productive approach is to study the difference between absolute plate motion (APM)

Fig. 8. (a) Orientation of finite strain for homogeneous and partitioned transpression. a is the angle of oblique convergence. The magnitude of finite strain is proportional to the amount of LPO and, therefore, to the amount of shear-wave splitting, (b) Partitioned deformation at any scale less than a seismic wave (e.g. shear bands) will tend to reduce the amount of splitting and the angle between the maximum splitting and the boundary. The correspondence between crustal deformation and shear-wave splitting (e.g. Little et al. 2002) indicates that this scale of deformation partitioning does not occur in the mantle.

MANTLE-DRIVEN DEFORMATION OF OROGENS

and relative plate motion (RPM) for plate margins with large transcurrent components of motion. The APM is fixed with respect to hotspots. The San Andreas system in central California and El Pilar system in Trinidad are a particularly useful comparison (Fig. 9), because of their difference in APM. Both margins record largely transcurrent motion and similar amounts of displacement (e.g. Crowell, 1966; Speed 1985). The San Andreas fault, defined as the plate boundary between the North American and the Pacific plates, moves southwestward in an absolute reference frame. Consequently, the San Andreas fault moves approximately perpendicular to its strike relative to the mantle (Figs 3 & 9). Essentially, the plate boundary between the North American and Pacific plate moves westward, overrunning its old position in the asthenosphere. In contrast, the plate boundary between the South American and Caribbean plates, reflected by the El Pilar fault of Venezuela and its equivalent in Trinidad, is fixed in an absolute reference frame (Figs 3 & 8). There is a discernible effect of APM on the deformation fabric, as reflected in the shearwave splitting observed in both places. The San Andreas system has low splitting values, with delay travel times only up to 1.25 s (Hartog & Schwartz 2001). In fact, models for interpreting shear-wave splitting in this area require a twolayer anisotropy model, in which only the upper layer reflects a fabric at a low angle to the San Andreas fault (e.g. Ozalaybey & Savage 1995; Hartog & Schwartz 2001). In general, within California, there is a gradual

59

rotation in the polarization of fast split shear waves from NE-SW in easternmost California to east-west in central California. This pattern can be interpreted as a large-scale drag in the mantle, reflecting a change from olivine LPO oriented parallel to the APM to that reflecting the San Andreas fault. Trinidad/Venezuela has relatively large splitting (Russo et al. 1996), with delay travel times of ~2 s. The large splitting indicates that both the lithospheric mantle and part of the asthenospheric mantle are pervasively deformed. The data also show alignment subparallel (~5-8° obliquity) to the Caribbean/South American plate boundary, unlike other transcurrent plate boundaries (San Andreas, New Zealand), where the splitting is at a higher angle to the major transcurrent faults. Because the Caribbean/ South American plate boundary is fixed in an absolute reference frame, we interpret the nearparallelism and high delay times to reflect the higher finite strains recorded along this boundary. Therefore, it appears that relative motion is not the only factor in the development of mantle fabric. Rather, there are also important structural implications and differences in mantle fabric resulting from the absolute plate motion. For instance, the Caribbean/South American plate boundary is a relatively narrow and welldefined boundary. There are essentially two plates that have relative movement and a ~ 100 km wide deformation zone between them. The strong fabrics and narrow deformational zone may result from the fixed position

Fig. 9. The role of absolute plate motion in mantle deformation, as shown by a comparison of the San Andreas and Trinidad systems. Both systems have approximately transcurrent relative motion. The San Andreas system moves WSW, with respect to a fixed mantle framework. Consequently, the lithospheric wrenching deformation (e.g. fault system) is moving with respect to the asthenosphere. This may result in low amounts of splitting and very heterogeneous deformation. In contrast, crustal movement in Trinidad is parallel to the absolute plate motion and the plate boundary does not move in a fixed mantle framework. The parallelism of shear-wave splitting and the narrowness (~100 km) of the deformation zone are a possible result of the lithospheric boundary being coincident with an asthenospheric boundary.

60

B. TIKOFFCTAL.

of the plate boundary in the mantle. In contrast, the Pacific/North America plate boundary in central California is wide and complicated. Instead of two plates, there is also the Sierra Nevada-Great Valley microplate that is moving to the northwest, although at a decreased rate. This complexity may result from the southwestward movement of the San Andreas fault system, in an absolute motion framework. Conclusions The similarity of deformation patterns in the upper crust (structural reconstruction, geodetic data, etc.) and upper mantle (shear-wave splitting, azimuthal anisotropy of Pn waves, etc.) indicate that these lithospheric layers are, at least, partially coupled. Crustal rocks in these deforming zones are hypothesized to be moving on a thin and flowing mantle, inferred from seismic attenuation studies. This deforming mantle controls orogenesis. Applying the clutch model - partial attachment between lithospheric layers - allows us to address three-dimensional models of convergent, transcurrent and divergent plate margins, for both crustal and mantle deformation. The history of the plate margin and the absolute plate motion are critical, as both influence the mantle fabric and overlying crustal structures.

AXEN, G.J., SELVERSTONE, J., BYRNE, T. & FLETCHER, J.M. 1998. If the strong crust leads, will the weak crust follow? GSA Today, 8, 1-8. BAI, Q., MACKWELL, S.J. & KOHLSTEDT, D.L. 1991. High-temperature creep of olivine single crystals, 1. Mechanical results for buffered samples. Journal of Geophysical Research, 96, 2441-2463. BARRUOL, G. & KERN, H. 1996. Seismic anisotropy and shear-wave splitting in lower-crustal and upper-mantle rocks from the Ivrea Zone: experimental and calculated data. In: MAINPRICE, D. & VAUCHEZ, A. (eds), Dynamics of the subcontinental mantle; from seismic anisotropy to mountain building. Physics of the Earth and Planetary Interiors, 95, 175-194. BARRUOL, G., HELFFRICH, G. & VAUCHEZ, A. 1997. Shear wave splitting around the northern Atlantic; frozen Pangaean lithospheric anisotropy? In: MAINPRICE, D., BOUDIER, F. & BOUCHEZ, J.L. (eds) The Adolphe Nicolas volume. Tectonophysics,279, 135-148. BEAVAN, J. & HAINES, J. 2001. Contemporary horizontal velocity and strain rate fields of the Pacific-Australian plate boundary zones through New Zealand. Journal of Geophysical Research, 106,741-770. BEN ISMAEL, W. & MAINPRICE, D. 1998. An olivine fabric database: an overview of upper mantle fabrics and seismic anisotropy. Tectonophysics, 296, 145-158. BORMANN, P., BURGHART, P.T., MAKEYEVA, L.I. &

VINNIK, L.P. 1993. Teleseismic shear-wave splitting and deformations in central Europe. Physics of the Earth and Planetary Interiors, 78, 157-166. BOSTOCK, M.G. & CASSIDY, J.F. 1995. Upper mantle BT and CT would like to thank the public library in Black stratigraphy beneath the southern Slave Craton. River Falls, Wisconsin, for providing a meeting room, in Canadian Journal of Earth Science, 34, 577-587. which the first version of this paper was outlined. The BRACE W.F. & KOHLSTEDT D.L. 1980. Limits on authors would like to thank Shun Karato for numerous lithospheric stress imposed by laboratory experconversations about olivine deformation. BT wishes to iments. In: HANKS, T.C. & RALEIGH, C.B. (eds), thank Scott Giorgis and Eric Horsman for work on the Magnitude of deviatoric stresses in the Earth's manuscript that allowed it to be submitted at all. crust and uppermost mantle. Journal of Geophysical Research, 85, 6248-6252. BURG, J.P., DAVY, P. & MARTINOD, J. 1994. ShortenReferences ing of analogue models of the continental lithosphere; new hypothesis for the formation of the ABRAMSON, E.H., BROWN, J.M., SLUTSKY, L.J. & Tibetan Plateau. Tectonics, 13, 475-483. ZAUG, J. 1997. The elastic constants of San Carlos Olivine to 17 GPa. Journal of Geophysical BYSTRIKY, M., KUNZE, K., BURLINI, L. & BURG, J.P. 2000. High shear strain of olivine aggregates: rheoResearch, 102, 12253-12263. logical and seismic consequences. Science, 290, ACHAUER, U., GLAHN, A., et al. 1994. New ideas on 1564-1567. the Kenya Rift based on the inversion of the combined dataset of the 1985 and 1989/90 seismic CALVERT, A., SANDVOL, E., et al. 2000. Geodynamic evolution of the lithosphere and upper mantle tomography experiments. In: PRODEHL, C., beneath the Alboran region of the western MediterKELLER, G.R. & KHAN, MA. (eds) Crustal and ranean: constraints from travel time tomography. upper mantle structure of the Kenya Rift. TectonoJournal of Geophysical Research, 105, 10871 — physics, 236, 305-329. 10898. AMATO, A., CHIARABBA, C. & SELVAGGI, G. 1997. Crustal and deep seismicity in Italy (30 years CARLSON, R.L. 1995. A plate cooling model relating rates of plate motion to the age of the lithosphere after). Annali di Geofisica, 40, 981-993. at trenches. Geophysical Research Letters, 22, ARMIJO, R., TAPPONNIER, P., MERCIER, J.-L. & HAN, 1977-1980. T.L. 1986. Quaternary extension in southern Tibet; field observations and tectonic implications. CARTER, N.L. & AVE LALLEMANT, H.G. 1970. High temperature flow of dunite and peridotite. GeologiJournal of Geophysical Research, 91, 13803 — cal Society of America Bulletin, 81, 2180-2202. 13872.

MANTLE-DRIVEN DEFORMATION OF OROGENS CHASTEL, Y.B., DAWSON, P.R., WENK, H.R. & BENNETT, K. 1993. Anisotropic convection with implications for the upper mantle. Journal of Geophysical Research, 98, 17757-17771. CHRISTENSEN, J.N., ROSENFELD, J.L. & DEPAOLA, D.J. 1989. Rates of tectonometamorphic processes from rubidium and strontium isotopes in garnet. Science, 244, 1465-1469. CHRISTENSEN, N.I. 2002. Continental mantle seismic anisotropy: a new look at the Twin Sisters massif. Tectonophysics, 355, 163-170. CHRISTENSEN, N.I. & MOONEY, W.D. 1995. Seismic velocity structure and composition of the continental crust; a global view. Journal of Geophysical Research, 100, 9761-9788. CROWELL, J.C. 1966. Displacements along the San Andreas fault, California. Annales Academia Scientiarum Fennicae, Series A3: GeologicaGeographica, 90, 119-125. DOSER, D.I. & YARWOOD, D.R. 1991. Strike-slip faulting in continental rifts; examples from Sabukia, East Africa (1928), and other regions. In: Gangi, A.F. (ed) World rift systems. Tectonophysics, 197, 213-224. DUNLAP, W.J., HIRTH, G. & TEYSSIER, C. 1997. Thermomechanical evolution of a ductile duplex. Tectonics, 16,983-1000. ENDERLE, U., MECHIE, J., SOBOLEV, S. & FUCHS, K. 1996. Seismic anisotropy within the uppermost mantle of southern Germany. Geophysical Journal International, 125, 747-767. FITCH, T.J. 1972. Plate convergence, transcurrent faults, and internal deformation adjacent to Southeast Asia and the Western Pacific. Journal of Geophysical Research, 77, 4432-4460. FOSSEN, H. & TIKOFF, B. 1993. The deformation matrix for simultaneous simple shearing, pure shearing and volume change, and its application to transpression-transtension tectonics. In: CASEY, M., DIETRICH, D., FORD, M., WATKINSON, J. & HUDLESTON, P.J. (eds) The geometry of naturally deformed rocks. Journal of Structural Geology, 15, 413-422. FOSSEN, H., TIKOFF, B. & TEYSSIER, C. 1994. Strain modeling of transpression and transtensional deformation. In: KRILL, A.G. & LIPPARD, S.J. (eds) Strike-slip tectonics: Proceedings of the 10th TSGS conference. Norsk Geologisk Tidsskrif, 74, 134-145. GAO, S., DAVIS, P.M., et al 1997. SKS splitting beneath continental rift zones. Journal of Geophysical Research, 102, 22781-22797. GIORGIS, S., MARKLEY, M., and TIKOFF, B. 2004. Vertical-axis rotation of rigid crustal blocks driven by mantle flow. In: GROCOT, J., MCCAFFREY, K.J.W., TAYLOR, G. & TIKOFF, B. (eds) Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 83-99. GOSCOMBE, B. 1992. High-grade reworking of central Australian granulites: Part 1, Structural evolution. Tectonophysics, 204, 361-399. HARTOG, R. & SCHWARTZ, S.Y. 2001. Depth-dependent mantle anisotropy below the San Andreas Fault system; apparent splitting parameters and wave-

61

forms. Journal of Geophysical Research, 160, 4155-4167. HATZFELD, D., KARAKOSTAS, V., et al. 2000. Microseismicity and faulting geometry in the Gulf of Corinth (Greece). Geophysical Journal International, 141, 438-456. HEARN T.M. 1996. Anisotropic Pn tomography in the Western United States. Journal of Geophysical Research, 101, 8403-8414. HEARN T.M. 1999. Uppermost mantle velocities and anisotropy beneath Europe. Journal of Geophysical Research, 104, 15123-15139. HEINTZ, M., VAUCHEZ, A., ASSUMPCAO, G., BARRUOL, G. & EGYDIO-SILVA, M. 2003. Shear wave spilling in SE Brazil: an effect of active on fossil upper mantle flow, or both? Earth and Planetary Science Letters, 211, 79-95. HELFFRICH, G. 1995. Lithospheric deformation inferred from teleseismic shear wave splitting observations in the UK. Journal of Geophysical Research, 100, 18195-18204. HERQUEL, G., WITTLINGER, G. & GUILBERT, J. 1995. Anisotropy and crustal thickness of NorthernTibet: new constraints for tectonic modelling. Geophysical Research Letters, 22, 1925-1928. HESS, H.H. 1964. Seismic anisotropy of the uppermost mantle under oceans. Nature, 203, 629-631. HOLE J.A., BEAUDOIN, B.C. & HENSTOCK, T.J. 1998. Wide-angle seismic constraints on the evolution of the deep San Andreas plate boundary by Mendocino triple junction migration. Tectonics, 17, 802-818. HOLT, W.E. 2000. Correlated crust and mantle strain fields in Tibet. Geology, 28, 67-70. IMBER, J., HOLDSWORTH, R.E. & BUTLER, C.A. 2001. A reappraisal of the Sibson-Scholz fault zone model: the nature of the frictional to viscous ('brittle-ductile') transition along a long-lived, crustal-scale fault, Outer Hebrides, Scotland. Tectonics, 20, 601-624. JAMES, D.E. & ASSUMPCAO, M. 1996. Tectonic implications of S-wave anisotropy beneath SE Brazil. Geophysical Journal International, 126, 1-10. JAMISON, W.R. 1991. Kinematics of compressional fold development in convergent wrench terranes. Tectonophysics, 190, 209-232. JORDAN, T.H. 1975. The continental tectosphere. Reviews of Geophysics and Space Physics, 13, 1-12. KAMINSKI, E. & RIBE, N.M. 2001. A kinematic model for recrystallization and texture development in olivine polycrystals. Earth and Planetary Science Letters, 189, 253-267. KENNEDY, L.A., RUSSEL, J.K. & KOPYLOVA, M.G. 2002. Mantle shear zones revisited: the connection between the cratons and mantle dynamics. Geology, 30, 419-422. KLOSKO, E., Wu, F., et al. 1999. Upper mantle flow in the New Zealand region from seismic anisotropy. Geophysical Research Letters, 26, 1497-1500. KOHLSTEDT, D.L., EVANS, B. & MACKWELL, S.J.

1995. Strength of the lithosphere: constraints

62

B. TIKOFFCTAL.

imposed by laboratory experiments. Journal of Geophysical Research, 100, 17587-17602. LAVE, J., AVOUAC, J.P., LACASSIN, R., TAPPONNIER, P. & MONTAGNER, J.P. 1996. Seismic anisotropy beneath Tibet; evidence for eastward extrusion of the Tibetan lithosphere? Earth and Planetary Science Letters, 140, 83-96. LAY, T. & WALLACE, T.C. 1995. Modern Global Seismology. Academic Press, San Diego, CA. LIN, S. & WILLIAMS, P.P. 1992. The geometrical relationship between the stretching lineation and the movement directions of shear zones. Journal of Structural Geology, 14, 491-497. LITTLE, T.A., SAVAGE, M.K. & TIKOFF, B. 2002. Relationship betweem crustal finite strain and seismic anisotropy in the mantle, Pacific Australia plate boundary zone, South Island, New Zealand. Geophysical Journal International, 151, 160-169. MACKWELL, S.J., KOHLSTEDT, D.L. & PATERSON, M.S. 1985. The role of water in the deformation of olivine single crystals. Journal of Geophysical Research, 90, 11319-11333. MARGHERITI, L., NOSTRO, C, AMATO, A. & Cocco, M. 1996. Seismic anisotropy; an original tool to understand the geodynamic evolution of the Italian Peninsula. In: MANTOVANI, E. (ed.) Geodynamics of the Mediterranean region and its implications for seismic and volcanic risk. Annali di Geofisica, 40, 759-769. MARSON-PIDGEON, K.A. & SAVAGE, M.K. 1997. Frequency-dependent anisotropy in Wellington, New Zealand. Geophysical Research Letters, 24, 3297-3300. MCKENZIE, D. & JACKSON, J. 1983. The relationship between strain rates, crustal thickening, palaeomagnetism, finite strain and fault movements within a deforming zone. Earth and Planetary Science Letters, 65, 182-202. MCNAMARA, D.E., OWENS, T.J., SILVER, P.O. & Wu, F.T. 1994. Shear wave anisotropy beneath the Tibetan Plateau. Journal of Geophysical Research, 99, 655-665. MELE, G., ROVELLI, A., SEBER, D., HEARN, T.M. & BARAZANGI, M. 1998. Compressional velocity structure and anisotropy in the uppermost mantle beneath Italy and surrounding regions. Journal of Geophysical Research, 103, 12529-12543. MOCHIZUKI, K., KASAHARA, J., SATO, T., SHINOHARA, M. & HIRATA, N. 1997. Seismic refraction tomography and its application in an experiment in the Tsushima Basin, Japan Sea. Geophysical Exploration, 50, 179-195. MOLNAR, P. 1992. Brace-Goetze strength-profiles, the partitioning of strike-slip and thrust faulting at zones of oblique convergence, and the stressheat flow paradox of the San Andreas fault. In: EVANS, B. & WONG, T.-F. (eds) Fault Mechanics and Transport Properties of Rocks. Academic Press, New York, 435-459. MOLNAR, P., ANDERSON, H.J., etal. 1999. Continuous deformation versus faulting through the continental lithosphere of New Zealand. Science, 286, 516519.

MONTAGNER, J.-P. & TANIMOTO, T. 1990. Global anisotropy in the upper mantle inferred from the regionalization of phase velocities. Journal of Geophysical Research, 95, 4797-4819. MONTAGNER, J.-P. & TANIMOTO, T. 1991. Global upper mantle tomography of seismic velocities and anisotropies. Journal of Geophysical Research, 96, 20337-20351. MOORE, M., ENGLAND, P. & PARSONS, B. 2002. Relation between surface velocity field and shearwave splitting in the South Island of New Zealand. Journal of Geophysical Research, 107, doi: 10.1029/2000JB000093. NISHIMURA, C.E. & FORSYTH, D.W. 1988. Rayleigh wave phase velocities in the Pacific with implications for azimuthal anisotropy and lateral heterogeneities. Journal of Geophysical Research, 94, 479-501. NISHIMURA, C.E. & FORSYTH, D.W. 1989. The anisotropic structure of the upper mantle in the Pacific. Geophysical Journal of the RAS, DGG, and EGS, 96,203-229. OKAL, E.A. & TALANDIER, J. 1997. T waves from the great 1994 Bolivian deep earthquake in relation to channeling of S wave energy up the slab. Journal of Geophysical Research, 102, 27421-27437. OLDOW, J.S., BALLY, A.W. & AVE LALLEMANT, H.G. 1990. Transpression, orogenic float, and lithospheric balance. Geology, 18, 991-994. OZALAYBEY, S. & SAVAGE, M.K. 1995. Shear-wave splitting beneath western United States in relation to plate tectonics. Journal of Geophysical Research, IW, 18135-18149. PHAKEY, P., DOLLINGER, G. & CHRISTIE, J. 1972. Transmission electron microscopy of experimentally deformed olivine crystals. In: Flow and fracture of rocks. Geophysical Monograph, 16, 117-138. PLOMEROVA, J., BABUSKA, V. & HORALEK, J. 1998. Seismic anisotropy and velocity variations in the mantle beneath the Saxothuringicum-Moldanubicum contact in Central Europe. In: PLOMEROVA, J., LIEBERMANN, R.C. & BABUSKA, V. (eds) Geodynamics of lithosphere and Earth's mantle: seismic anisotropy as a record of the past and present dynamic processes. Pure and Applied Geophysics, 151, 365-394. PLOMEROVA, J., KOUBA, D. & BABUSKA, V. 2002. Mapping the lithosphere-asthenosphere boundary through changes in surface-wave anisotropy. Tectonophysics, 358, 175-186. RAITT, R.W., SHOR, G.G., FRANCIS, T.J.G. & MORRIS, G.B. 1969. Anisotropy of the Pacific upper mantle. Journal of Geophysical Research, 74, 3095-3109. REY, P., VANDERHAEGHE, O. & TEYSSIER, C. 2002. Gravitational collapse of continental crust: definitions, regimes, mechanisms and modes. Tectonophysics, 342, 435-449. RIBE, N.M. & Yu, Y. 1991. A theory for plastic deformation and textural evolution of olivine polycrystals. Journal of Geophysical Research, 96, 8325-8335. RICHARD, P. & COBBOLD, P. 1990. Experimental insights into partitioning fault motions in continen-

MANTLE-DRIVEN DEFORMATION OF OROGENS tal convergent wrench zones. In: BOCCALETTI, M. & NUR, A. (eds) Active and recent strike-slip tectonics; selected papers from an international workshop. Annales Tectonicae, 4, 35-44. ROTH, E.G., WIENS-DOUGLAS, A., DORMAN, L.M., HILDEBRAND, J. & WEBB, S.C. 1999. Seismic attenuation tomography of the Tonga-Fiji region using phase pair methods. Journal of Geophysical Research, 104, 4795-4809. ROYDEN,

L.H., BURCHFIEL,

B.C., KlNG, R.W.,

WANG, E., CHEN, Z., SHEN, F. & Liu, Y. 1997. Surface deformation and lower crustal flow in eastern Tibet. Science, 276, 788-790. Russo, R.M. & OKAL, E.A. 1998. Shear wave splitting and upper mantle deformation in French Polynesia: evidence from small-scale heterogeneity related to the Society Hotspot. Journal of Geophysical Research, 103, 15089-15107. Russo, R.M., SILVER, P.G., FRANKE, M., AMBEH, W.G. & JAMES, D.E. 1996. Shear-wave splitting in northeast Venezuela, Trinidad, and the eastern Caribbean. Physics of the Earth and Planetary Interiors, 95, 251-275. SANDVOL, E., Ni, J., OZALAYBEY, S. & SCHLUE, J. 1992. Shear-wave splitting in the Rio Grande Rift. Geophysical Research Letters, 19, 23372340. SAVAGE, M.K. 1999. Seismic anisotropy and mantle deformation; what have we learned from shear wave splitting? Reviews of Geophysics, 37, 65-106. SAVAGE, M.K. & SILVER, P.G. 1993. Mantle deformation and tectonics; constraints from seismic anisotropy in the western United States. In: MAINPRICE, D., VAUCHEZ, A. & MONTAGNER, J.P. (eds) Dynamics of the subcontinental mantle: from seismic anisotropy to mountain building. Physics of the Earth and Planetary Interiors, 78, 207-227. SCHOLZ, C.H. 1988. The brittle-ductile transition and the depth of seismic faulting. Geologisches Rundschau, 11, 319-328. SIB SON. R.H. 1977. Fault rocks and fault mechanisms. Journal of the Geological Society of London, 133, 191-213. SILVER, P.G. 1996. Seismic anisotropy beneath the continents: probing the depths of geology. Annual Review of Earth and Planetary Sciences, 24, 385-432. SILVER, P.G. & CHAN, W.W. 1991. Shear wave splitting and subcontinental mantle deformation. Journal of Geophysical Research, 96, 1642916454. SILVER, P.G. & HOLT, W.E. 2002. The mantle flow field beneath western North America. Science, 295, 1054-1058. SPEED, R.C. 1985. Cenozoic collision of the Lesser Antilles Arc and continental South America and the origin of the El Pilar fault. Tectonics, 4, 41 -69. STERN, T.A. & MCBRIDE, J.H. 1997. Seismic exploration of continental strike-slip zones. In: KLEMPERER, S.L. & MOONEY, W.D. (eds) Deep seismic profiling of the continents: I, General

63

results and new methods. Tectonophysics, 286, 63-78. TAPPONNIER, P., ZHIQIN, X., ROGER, F., MEYER, B., ARNAUD, N., WITTLINGER, G. & JINGSUI, Y. 2002. Oblique stepwise rise and growth of the Tibetan Plateau. Science, 294, 1671-1677. TEYSSIER, C. & TIKOFF, B. 1998. Strike-slip partitioned transpression of the San Andreas fault system: a lithospheric scale approach. In: HOLDSWORTH, R.E., STRACHAN, R.A. & DEWEY, J.F. Continental Transpression and Transtension Tectonics, (eds) Geological Society, London, Special Publications, 135, 143-158. TEYSSIER, C., TIKOFF, B. & MARKLEY, M. 1995. Oblique plate motion and continental tectonics. Geology, 23, 447-450. TIKOFF, B. & GREENE, D. 1997. Stretching lineations in transpressional shear zones; an example from the Sierra Nevada Batholith, California. Journal of Structural Geology, 19, 29-39. TIKOFF, B., TEYSSIER, C. & WATERS, C.L. 2002. Clutch tectonics and partial attachment of lithospheric layers. EUG Special Publication Series, 1,93-117. TOMMASI, A. & VAUCHEZ, A. 2001. Continental rifting parallel to ancient collisional belts: an effect of the mechanical anisotropy of the lithospheric mantle. Earth and Planetary Science Letters, 185, 199-210. TOMMASI, A., TIKOFF, B. & VAUCHEZ, A. 1999. Upper mantle tectonics: three-dimensional deformation, olivine crystallographic fabrics and seismic properties. Earth and Planetary Science Letters, 168, 173-186. TOMMASI, A., MAINPRICE, D., CANOVA, G. & CHASTEL, Y. 2000. Viscoplastic self-consistent and equilibrium-based modeling of olivine lattice preferred orientations: implications for the upper mantle seismic anisotropy. Journal of Geophysical Research, 105, 7893-7908. VANDERHAEGHE, O. & TEYSSIER, C. 2001. Crustal scale Theological transitions during late-orogenic collapse. Tectonophysics, 335, 211-228. VAN DER LEE, S. & NOLET, G. 1997. Seismic image of the subducted trailing fragments of the Farallon Plate. Nature, 386, 266-269. VAUCHEZ, A. & GARRIDO, C.J. 2001. Seismic properties of an asthenospherized lithospheric mantle: constraints from lattice preferred orientations in peridotite from the Ronda Massif. Earth and Planetary Science Letters, 192, 235-249. VAUCHEZ, A., TOMMASI, A. & EGYDIO-SANTOS, M. 1994. Self-indentation of a heterogeneous continental lithosphere. Geology, 22, 967-970. VAUCHEZ, A., BARRUOL, G. & TOMMASI, A. 1997. Why do continents break-up parallel to ancient orogenic belts? Terra Nova, 9, 62-66. VINNIK, L.P., FARRA, V. & ROMANOWICZ, B. 1989a. Azimuthal anisotropy in the Earth from observations of SKS at GEOSCOPE and NARS broadband stations. Bulletin of the Seismological Society of America, 79, 1542-1558. VINNIK, L.P., KIND, R., KOSAREV, G.L. & MAKEYEVA, L.I. 1989&. Azimuthal anisotropy in the lithosphere

64

B. TIKOFFETAL.

from observations of long-period S-waves. Geophysics Journal International, 99, 549-559. VINNIK, L.P., KRISHNA, V.G., KIND, R., BORMANN, P. & STAMMLER, K. 1994. Shear wave splitting in the records of the German Regional Seismic Network. Geophysical Research Letters, 21, 457-460. WENK, H.R. & TOME, C.N. 1999. Modeling dynamic recrystallization of olivine aggregates deformed in simple shear. Journal of Geophysical Research, 104, 25513-25527. WENK, H.R., BENNETT, K., CANOVA, G.R. & MOLINARI, A. 1991. Modelling plastic deformation of peridotite with the self-consistent theory. Journal of Geophysical Research, 96, 83378349. WITHJACK, M.O. & JAMISON, W.R. 1986. Deformation produced by oblique rifting. Tectonophysics, 126, 99-124.

WOLFE, C.J. & SILVER, P.G. 1998. Seismic anisotropy of oceanic upper mantle; shear wave splitting methodologies and observations. Journal of Geophysical Research, 103, 749-771. WOLFE, C.J. & VERNON, F.L. 1998. Shear-wave splitting at central Tien Shan: evidence for rapid variation of anisotropic patterns. Geophysical Research Letters, 25, 1217-1220. YIN, A., KAPP, P.A., et al. 1999. Significant late Neogene east-west extension in northern Tibet. Geology, 27, 787-790. ZHANG, S. & KARATO, S. 1995. Lattice preferred orientation of olivine aggregates in simple shear. Nature, 375, 774-777. ZHANG, S., KARATO, S., FITZGERALD, J., PAUL, U. & ZHOU, Y. 2000. Simple shear deformation of olivine aggregates. Tectonophysics, 316, 133-152.

Termination of strike-slip faults at convergence zones within continental transform boundaries: examples from the California Continental Borderland M. R. LEGG1, M. J. KAMERLING2 & R. D. FRANCIS3 l

Legg Geophysical, Huntington Beach, CA 92647, USA (e-mail: [email protected])

^Institute for Crustal Studies, University of California, Santa Barbara, CA 93106, USA 3

Department of Geological Sciences, California State University Long Beach, CA 90840, USA Abstract: Continental transform plate boundaries are broad, composed of numerous active and subparallel strike-slip fault zones. Irregular geometry along the major transform structure creates convergence and divergence zones within the plate boundary where other strike-slip faults terminate. Some prominent irregularities result from microplate interactions. Relative fault displacement, diminishing to zero at fault terminations, must be accommodated or transferred to other structures, laterally or vertically, away from the fault end-point. Distinct styles of strike-slip fault termination may represent different degrees of vertical strain partitioning within the plate boundary. The Western Transverse Ranges (WTR) of California mark a major structural discontinuity that cuts at high angle across the Pacific-North America transform boundary. Within the California Continental Borderland, two end-member classes of right-slip fault termination against the WTR are apparent. (1) Several major faults, including the San Clemente, San Pedro Basin, Ferrelo and Newport-Ingle wood, intersect the southern boundary of the WTR at high angles, with negligible to minor local deflection and minor dissipation of right shear at the Earth's surface. These faults are inferred to cut through the entire borderland crust and continue in the lower crust beneath the WTR, as evident in geophysical data. These 'blind' near vertical faults may control segmentation and earthquake activity on the overlying west-trending WTR structures. (2) In contrast, the Palos Verdes and possibly Whittier faults appear broadly deflected westward to merge at low angle with WTR structure. NW-trending faults rotate counterclockwise, away from the axis of principal shortening as observed in pure shear models, and slip is dissipated through folding, thrust transfer and rotation. Deflected faults are inferred to be predominately upper crustal features, detached from the lower crust and unable to underthrust the WTR. These two distinct right-slip fault termination styles, and associated convergent structures, suggest that basal shear drives vertical-axis rotation of the WTR block over the underthrust Inner Borderland plate. Furthermore, the lower plate, slivered by these right-slip faults, is incompletely coupled with the Pacific plate.

Broad continental transform plate boundaries are often composed of numerous active and generally subparallel strike-slip fault zones. Geometrical irregularities along the transform zone, along with complex multiple plate interactions, may result in local collision zones as well as detached crustal microplates. These irregular structures limit the lateral and vertical extent of strike-slip fault zones. Terminations of strike-slip faults against collision zones, detached microplates and other local structures have important implications for the history and character of a transform plate

boundary. Relative strike-slip fault displacement diminishes toward fault terminations and must be accommodated or transferred to other structures, laterally or vertically, away from the fault endpoint. This results in termination features such as horsetail splays, termination bulges, restraining bends, rotated blocks and pull-apart basins (Fig. 1; described by Freund 1971; ChristieBlick & Biddle 1985; Harding et al. 1985; among others). These distinct termination styles may represent different degrees and types of strain partitioning within the plate boundary.

From: GROCOTT, J., MCCAFFREY, K. J. W., TAYLOR, G. & TIKOFF, B. (eds) 2004. Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 65-82. 0305-8719/04/$15 © The Geological Society of London 2004.

66

M. R. LEGGETAL.

Fig. 1. Diagram showing termination styles of strike-slip faults. For transform faults, termination occurs at another plate boundary such as a spreading centre, subduction zone, or another transform fault, sometimes forming a triple junction. Where a strike-slip fault ends at convergent boundaries, such as subduction zones or as a tear fault in fold-and-thrust belts, the underthrust edge of the plate or block continues beyond the surface point of strike-slip fault termination (dotted lines).

A transform boundary is partitioned, vertically or horizontally, into blocks that interact with one another to accommodate strain between the plates. For example, rotated or translated blocks may be affected primarily along their edges, but if they are detached from the lower crust, basal shear may be a more important factor in their movements. Some termination styles, such as those against subduction zones, have straight surface traces and indicate continuation in the underthrust plate beneath the detached upper plate (Fig. 1). Where straight fault traces terminate in areas of block rotation, detachment of the rotating block is required, and space problems for the restraining step-over geometry shown force thrusting of the rotating block edges over adjoining blocks or internal deformation of the blocks. Others, such as horsetail splays and termination bulges, could indicate faults that terminate entirely, without subsurface continuation, and relative slip is absorbed by distributed deformation within the fault-bounded blocks. The termination style of a fault is thus closely related to its vertical extent, and to the story it tells about plate boundary dynamics.

Pacific-North America transform plate boundary The Pacific-North America transform plate boundary in southern California is an excellent laboratory for studying strike-slip fault terminations. The plate boundary is a zone up to 400 km wide, from the eastern edge marked by

the San Andreas fault to the westernmost active fault zones within the offshore California Continental Borderland (Fig. 2). A major left bend in the southern San Andreas fault between the Salton Trough and the central California Coast Ranges creates a restraining geometry (Crowell 1974) that impedes smooth NW translation of the Pacific plate edge past North America. Consequently, oblique convergence (transpression) in the southern California region increases the breadth of the active transform boundary and the number of active structures, compared to more smoothly linear and favourably oriented transform fault zones, such as that to the north of the Western Transverse Ranges. NW slip in the transform boundary is further impeded by the presence, entirely within the boundary zone, of the Western Transverse Ranges (WTR), an elongate 250-300 km long by 75-90 km wide block with N65°W-trending northern and southern boundaries (Figs 2 & 3). This block, which includes several east-west mountain ranges and basins, is oriented at high angle, about 50 degrees, to the relative motion vector of the Pacific-North America transform boundary (Fig. 2). This impediment further enhances the oblique convergence in southern California caused by the curved San Andreas fault and spreads this convergence over a broad area. The WTR played a critical role in the evolution of the plate boundary and of the California Continental Borderland. It has undergone about 90 degrees of vertical-axis clockwise rotation in the last 17-18 Ma, after rifting away from the southern California coast (Kamerling &

TERMINATION OF STRIKE-SLIP FAULTS

Fig. 2. Map showing major faults and tectonic blocks of southern California-northern Baja California (modified from Lonsdale 1991). Oceanic crust of the Pacific plate occurs west of the Patton Escarpment. Continental crust of the North American plate that is relatively unaffected by transform movement occurs east of the San Andreas fault and the Gulf of California. Between these two regions is the Pacific-North America transform plate boundary, which is very broad in southern California due to the major left bend in the San Andreas fault zone. The continental margin has been rifted obliquely in two locations: the Gulf of California and the Inner California Continental Borderland. Remnant pieces (microplates) of the former Farallon plate were captured by the Pacific plate and influenced the subsequent plate boundary evolution. NW-trending right-slip faults of the California Continental Borderland terminate against the Western Transverse Ranges, a block defined by N65°W-trending faults along its northern and southern boundaries.

67

68

M. R. LEGGETAL

Fig. 3. Map showing major faults and seismicity in the southern California region near the intersection of the NW-trending California Continental Borderland and Peninsular Ranges structure with the west-trending Western Transverse Ranges structure. Location of Figures 8, 9 and 10 are shown for reference. CuFZ, Cucamonga fault zone; NIFZ, Newport-Ingle wood fault zone; PVF, Palos Verdes fault zone; RF, Raymond fault; SGaFZ, San Gabriel fault zone; SMaFZ, Sierra Madre fault zone; SMoFZ, Santa Monica-Holly wood fault zone; CH, Chino Hills; LA, Los Angeles; SBA, Santa Barbara; SD, San Diego.

Luyendyk 1979, 1985; Hornafius et al 1986; Legg 1991; Luyendyk 1991; Bohannon & Geist 1998). Palaeomagnetic data suggest that the WTR rotated as an elongate rigid block, although some internal deformation is evident, especially left-slip on NE- to east-trending faults as well as folding and thrust faulting that occurred during the later time periods when the block became highly oblique to the plate motion vector. Various models address coupling of the WTR with other blocks. Luyendyk et al (1980) consider that edge interaction among elongate rigid blocks caused the WTR to rotate around a pivot point (Fig. 4a), while Nicholson et al. (1994) invoke basal traction resulting from the capture of the Monterey microplate by the Pacific plate (Fig. 4b). The basal traction model implies that the WTR is detached from the underlying Monterey microplate, and predicts that the decollement is a low-angle transform fault. Thus the boundary between the Pacific plate and the

North America plate is irregular in three dimensions (Fig. 4) and its configuration changes with time as the plate boundary evolves. Along the southern boundary of the WTR, major, NW-trending, strike-slip fault zones terminate along a west-trending fold-and-thrust belt. This paper discusses termination styles of these faults, and how these provide important clues about plate interaction and strain partitioning, including relative contributions of edge-driven and basal traction-driven vertical-axis block rotations.

Strike-slip fault termination styles Termination styles of right-slip faults against the southern edge of the WTR are evidenced by surface and seismic mapping, oil field data, seismicity data, aeromagnetic data and other information. On a gross scale, two end-member styles are apparent: (1) straight fault termination; (2) bending fault termination. Some localized

TERMINATION OF STRIKE-SLIP FAULTS

69

NW-trending right-slip fault zones of the southern California region may be classified into one of these two basic termination styles.

Fig. 4. Diagrams showing two mechanisms for clockwise vertical-axis rotation of the Western Transverse Ranges (WTR) crustal block. Note that both models require a detached WTR block, (a) The edge-driven model first proposed by Luyendyk et al. (1980) pushes the WTR due to increasing NW movement of the adjacent blocks bounded by NW-trending right-slip faults. The primary transform fault is the high-angle San Andreas fault at present. If the NW movement of blocks does not match the rotation rate of the WTR, then deformation of the blocks must occur, resulting in various forms of strike-slip fault termination, (b) Basal traction as a driving force is implied by the model proposed by Nicholson et al. (1994) wherein Pacific plate capture of the Monterey microplate along an irregular break in the North America continental margin forced the clockwise vertical-axis rotation of the Western Transverse Ranges (WTR). In this model, both high-angle and low-angle transform faults are created. Two cases for microplate capture are proposed, although in reality a sheared microplate may have evolved from an originally intact microplate as right-lateral shear progressed.

bending of surficial faults may exist at straight fault terminations, and other combinations of these two basic styles may occur at other terminations. Notwithstanding, most of the major

Straight fault termination Straight fault termination is defined by the continuation of the major strike-slip fault traces, or principal displacement zone, in an essentially straight line directly up to the intersection with the major cross-cutting structure. For a crustal thickness of 20-30 km in the southern California borderland (Shor & Raitt 1958; ten Brink et al. 2000), a few kilometres of local bending of the surface fault trace before termination likely results from shallow crustal deformation, often within the weak sedimentary layers of strikeslip basins. Yet the major crustal fault zone is considered to have a straight fault termination style. Convergence of material on the SW side of the right-slip faults must be accommodated at the fault termination against the WTR. If the net slip diminishes to zero at the right-slip fault termination and intersection with the cross-cutting structure, excess material may be accommodated in folding or crustal thickening (Fig. 1). Alternatively, the material to the SW may be underthrust beneath the WTR structural boundary and right-slip is non-zero at depth, with only local shallow folding to accommodate the surface dissipation of right-slip (Figs 4b & 5). A third way to accommodate incoming material involves vertical-axis rotation of crustal blocks, as proposed by Luyendyk et al. (1980) for the WTR (Fig. 4a). In this situation, the right-slip must exactly equal the net NW component of WTR block motion, due to clockwise rotation, at the distance of the fault intersection from the pivot point of rotation. A consequence of this rigid block rotation model is that triangular basins form at the NE side of the right-slip fault termination. If the right-slip exceeds the net NW motion of the rotating block at the intersection, then a triangular fold-and-thrust zone should form, like an accretionary wedge in a subduction zone with a diminishing convergence rate away from the intersection. Generally, the block rotations are confined to rigid upper crustal blocks that must be detached from deeper, possibly more ductile, shear deformation in the lower plate. For real crustal faults, combinations of the three mass conservation mechanisms occur with varying degrees of each mechanism acting depending on the strength of coupling, geometric configuration, pre-existing structural fabric, or other heterogeneities in the crustal rheology.

70

M. R. LEGG ETAL.

Fig. 5. Model showing the Inner Borderland fault blocks as the lower plate is imderthrust beneath the clockwise-rotating Western Transverse Ranges block. Dextral shear couples at the base of the WTR (case 2 in Fig. 4b) impose a clockwise torque on the overriding block. Surficial sediments are off-scraped against the major thrust and oblique-reverse faults at the southern boundary of the WTR; lengths of arrows show that convergence rates diminish with distance away from each major right-slip fault termination due to the continuing clockwise rotation of the WTR. Inset lists some of the major complications ignored for this simple model.

San Clemente fault system. The San Clemente fault system comprises a long, more than 590 km, and well-defined zone of dextral shear in the California Continental Borderland (Legg 1985). At the seafloor, it terminates against the WTR as the Santa Cruz-Catalina Ridge fault zone near the eastern end of Santa Cruz Island (Fig. 3). The principal displacement zone generally demarcates the southwestern margin of Santa Monica Basin along a straight trace striking N50°W. Locally, within about 2-3 km of the fault termination, a slight counterclockwise deviation from this trend is mapped (Junger & Wagner 1977), but this may result from very shallow deformation within the Late Cenozoic sedimentary rocks of the borderland along the elevated ridge, or from local changes in fault trend typical of active strike-slip faults (e.g. Freund 1971; Christie-Blick & Biddle 1985). Uplift of the Santa Cruz-Catalina Ridge may be due to transpression, although the 1981 Santa Barbara Island earthquake near the southern end of the ridge showed nearly pure right-slip (Corbett

1984). Transpression along the Santa Cruz-Catalina Ridge, and horsetail splaying of the San Clemente fault system between Santa Barbara Island and Santa Cruz Island, may accommodate some diminishing right-slip on the SW side of this major shear zone. Other evidence, however, suggests continuation of the San Clemente fault system to the NW, beneath Santa Cruz Island and the Santa Barbara Channel. Geophysical data including gravity and magnetic anomalies (Fig. 6) show a major ridge (positive anomalies) parallel to and along the southwestern flank of the San Clemente fault system (Langenheim et al. 1993). Even though the exact source of the anomaly ridge is uncertain, it is clearly associated with the edge of the major crustal block bounded by the northern Santa Cruz-Catalina Ridge fault zone. Furthermore, the geophysical anomaly continues to the NW beyond Santa Cruz Island and into the Santa Barbara Channel, implying that the edge of the crustal block and the Santa Cruz-Catalina Ridge fault zone continue to the NW beyond the

TERMINATION OF STRIKE-SLIP FAULTS

Fig. 6. Aeromagnetic map of the California Continental Borderland showing the NW-trending anomaly ridge west of the San Clemente fault system (after Langenheim et al. 1993). Major faults are shown as dark solid and dashed lines, and the coast and island outlines as thick black lines. The magnetic ridge is likely due to shallow volcanic rocks, ophiolite, or intrusives along the eastern edge of the Outer Borderland block. The ridge continues to the NW beyond Santa Cruz Island and the southern boundary of the WTR, suggesting that the San Clemente fault system also continues as a blind strike-slip fault beneath the Santa Barbara Channel. Abbreviations as in Figure 3.

71

72

M. R. LEGGETAL.

southern boundary of the WTR. Thus, the Inner Borderland crust along the San Clemente fault system is thrust beneath the WTR including the major dextral shear represented at the surface by the Santa Cruz-Catalina Ridge fault zone (Fig. 5), as would occur at a subduction zone. This thrust, inferred to be the Channel Islands thrust, appears in exploration industry multichannel seismic reflection profiles located to the south of Santa Cruz Island (Fig. 7). Folding of material in the hanging wall of the north-dipping thrust represents off-scraping of the shallow sedimentary strata like an accretionary wedge of a subduction zone. The shortening, by folding and thrust faulting, appears to diminish to the west, however, along the south flank of the Northern Channel Islands platform (lunger 1979), forming a triangular-shaped accretionary wedge in map view (Fig. 8). In contrast, a small, triangular-shaped, pull-apart(?) basin occurs on the opposite, NE side of the Santa Cruz-Catalina Ridge fault zone (Junger & Wagner 1977). These secondary structures accommodate some part of the right-slip discontinuity across the Santa Cruz-Catalina Ridge fault zone at its

termination. The dissipation of deformation with distance from the right-slip fault implies that vertical-axis clockwise rotation of the WTR continued through Late Cenozoic time as observed in palaeomagnetic declinations (Kamerling & Luyendyk 1979, 1985; Luyendyketal. 1980). If the San Clemente fault system continues beneath the Santa Barbara Channel, then it may be considered a 'blind' strike-slip fault system. Continued dextral shear on the subducted shear zone should affect the structure and deformation within the hanging wall of the Channel Islands thrust. Evidence suggesting this interaction appears in the microseismicity distribution (Fig. 8), where dense clusters of earthquake epicentres lie to the NE of the projected blind rightslip fault trace, and a broad quiet area with few epicentres exists to the SW of the fault trace. This pattern of subdued seismicity to the west may represent increased convergent base shear that locks shallow faulting or possibly is a stress shadow resulting from the 21 December 1812 Santa Barbara earthquake. In contrast, the possible dilational basal shear to the east could enhance brittle failure and seismicity. Based on

Fig. 7. Multichannel seismic reflection profile that shows the major thrust fault located south of Santa Cruz Island. This fault is inferred to be the Channel Islands thrust along which the Inner Borderland is underthrust beneath the WTR. Fault imaged in seismic profile cuts through Late Cenozoic sedimentary and volcanic rocks of the Inner Borderland and appears to have another thrust, and possible oblique-slip high-angle faults located in the hanging wall.

TERMINATION OF STRIKE-SLIP FAULTS

73

Fig. 8. Map showing major faults and seismicity in the Santa Barbara Channel region. Epicentres are from the Southern California Seismograph Network and include events with M > 3.0 for the period 1932-2000 plus year 2001 events shown in Figure 3. Faults and folds (lines with tick marks) south of Santa Cruz Island are from lunger (1979) and show the triangular-shaped convergent region to the west and pull-apart basin to the east of the Santa CruzCatalina Ridge fault zone termination. The dashed grey line shows the inferred trace of the blind San Clemente fault system beneath the Santa Barbara Channel, which appears to separate the aseismic western Santa Barbara Channel area from a more active eastern area. Bathymetric contours also appear to follow this trend (contours in metres).

the epicentre pattern, the interaction may persist to the NW as far as the coastline near Santa Barbara. Bathymetry in the Santa Barbara Channel also supports the blind right-slip fault interpretation; a shoaler eastern Santa Barbara Channel is separated by NW-trending contours, parallel to the projected fault trace, from the deeper central Santa Barbara Basin (Fig. 8). Furthermore, west-trending thrust or oblique-reverse faults in the upper crust of the Santa Barbara Channel seem to be broadly deflected, in a right-lateral sense, across the projected San Clemente fault system (Vedder et al 1986; Jennings 1994). San Pedro Basin fault zone. A second major right-slip fault zone offshore of southern California that shows a straight fault termination style is the San Pedro Basin fault zone (Figs 3 & 9). Prominent seafloor deformation occurs along

the San Pedro Basin fault zone where it crosses the lower slope of Santa Monica Bay on a linear N50°W trend directly toward Point Dume and the Dume submarine canyon. Indeed, the intersection of the NW-trending San Pedro Basin fault with the west-trending AnacapaDume reverse fault occurs at the mouth of the Dume canyon where it discharges into the deeper Santa Monica Basin. Point Dume and the canyon location may be a direct result of the differential shortening imparted by the intersection of the NW-trending dextral San Pedro Basin fault zone. Evidence that the San Pedro Basin fault zone underthrusts the WTR (Santa Monica Mountains) appears as the change in character of the Anacapa-Dume fault across the intersection (Fig. 9). The strike of the Anacapa-Dume fault and the trend and height of the seafloor escarpment change at the intersection. A buried domal uplift is also mapped in the

74

M. R. LEGGETAL.

Fig. 9. Map showing the WTR-Inner Borderland boundary in the coastal Los Angeles region. Shaded relief bathymetry and topography, from the US Geological Survey (Dartneil & Gardner 1999), show tectonic uplifts along the active Newport-Ingle wood and San Pedro Basin fault zones. Major faults are labelled with italic type. Fault locations are simplified from Richmond et al (1981), Burdick & Richmond (1982), Jennings (1994), Francis et al. (1999), Tsutsumi et al (2001) and Shaw et al (2002) for the north Los Angeles Basin area. Major anticlinoria located west of the Palos Verdes fault zone include the Shelf Projection Anticlinorium, Redondo Platform Anticlinorium, Palos Verdes Hills Anticlinorium and San Pedro Bay Anticlinorium (Nardin & Henyey 1978). WBHL, West Beverly Hills Lineament.

hanging wall of the thrust, west of the intersection (Sorlien et al. 2001). The thicker crust associated with the uplifted Santa Monica Mountains may obscure other geophysical signals associated with a blind San Pedro Basin fault zone. Newport-Inglewoodfault zone. The NewportIngle wood fault zone, another high-angle rightslip fault zone, appears to continue with a straight trace to the intersection with the southern boundary of the WTR, at the Santa MonicaHollywood fault zone in the northwestern Los Angeles basin (Figs 9 & 10; Wright 1991; Tsutsumi et al. 2001). The northern Newport-Inglewood fault zone, like the San Pedro Basin fault zone, consists of discontinuous and en echelon fault segments with prominent geomorphic

expression across the southwestern margin of the Los Angeles basin. Transpression, generally associated with restraining bends and fault stepovers, forms the prominent uplifts along the Newport-Inglewood fault zone that hold prolific petroleum reservoirs (Barrows 1974; Wright 1991). The subsurface information from petroleum exploration and production in the region provides a comprehensive image of the shallow crustal structure of the Newport-Inglewood fault zone. At its northern end, the Newport-Inglewood fault zone appears to continue at the surface as the West Beverly Hills lineament (Wright 1991; Tsutsumi et al. 2001) into the Santa Monica Mountains and offsets the Santa MonicaHollywood fault zone with a left-lateral separation.

TERMINATION OF STRIKE-SLIP FAULTS

75

Fig. 10. Map showing major faults and recent seismicity in the Los Angeles region. Data sources are as in Figures 3, 6 and 9 plus blind faults and oil field locations from Tsutsumi et al. (2001). Earthquake sequences associated with the Newport-Inglewood fault zone (NIFZ) occurred on 9 September 2001 and 28 October 2001 (Hauksson et al. 2002). The northern sequence (9 September) is located in the vicinity of the West Beverly Hills lineament (WHBL; Tsutsumi et al. 2001), which appears to offset the west-trending southern boundary of the WTR, between the Santa Monica and Hollywood fault zones. The locations of the 1971 San Fernando and 1994 Northridge earthquake mainshocks are shown. Blind thrust faults and oil fields at the northern termination of the NIFZ are related to shortening and underthrusting of the Inner Borderland beneath the WTR; the Quaternary Hollywood Basin (HB) is considered a pull-apart basin associated with the differential shortening due to right-slip across the NIFZ.

The northern continuation of the Inglewood fault past the Santa Monica fault cannot be traced with available data, due to complex structural relations in that area (Wright 1991; Tsutsumi et al. 2001). But there is a pronounced structural discontinuity apparent, involving complexly folded and faulted Late Cenozoic sediments and sedimentary rocks in the oil fields of that area. Immediately to the east, and south of the range frontal Hollywood fault, lies the Hollywood Basin, which contains more than 300 m of Mid to Late Quaternary strata and is inferred to be a pull-apart basin (Tsutsumi et al. 2001). Right-slip on the Inglewood fault diminishes rapidly northward from the Inglewood oil field as it approaches the Santa Monica-Holly wood fault zone and the WTR, with some of this slip being accommodated in

the Cheviot Hills and other fold-and-thrust structures on the SW side of the termination. The paired triangular fold belt to the SW of the Inglewood fault termination and the pull-apart Hollywood Basin to the NE are considered equivalent to the termination structures at the northern end of the shallow San Clemente fault system at Santa Cruz Island. The decrease in shortening to the west and extension to the east of the strike-slip fault termination confirm that clockwise rotation of the WTR block continued through Quaternary time. Otherwise, the relative shortening/extension would be more uniform along, and the structural axis would be parallel to, the southern boundary of the WTR - at least until the next right-slip fault intersection is reached. Furthermore, assuming a rigid WTR block, if there were no vertical-axis rotation of

76

M. R. LEGGCTAL.

the WTR in Quaternary time, then the Quaternary shortening rates should increase systematically westward as each right-slip fault intersection is crossed. The shortening rates, although still uncertain in detail, appear to show greater shortening rates to the east, along the Cucamonga and Sierra Madre faults (WGCEP 1995), with slower shortening across the Santa Monica, Hollywood and western thrust faults at the southern WTR boundary (Tsutsumi et al 2001). Additional shortening accommodated within the WTR block, particularly from the San Fernando Valley westward to the Santa Barbara Channel, involves non-rigid behaviour across the short axis of the WTR block. Notwithstanding, the southern boundary of the WTR is remarkably straight, suggesting rigid, vertical-axis, rotation as a coherent block, consistent with the palaeomagnetic data (Kamerling & Luyendyk 1979, 1985; Luyendyketal. 1985). Evidence that the Newport-Ingle wood fault zone continues as a blind fault beneath the WTR (Santa Monica Mountains) appears in careful relocations of seismicity near the intersection of the cross-cutting structures (Fig. 10; Hauksson et al. 2002). A sequence of micro-earthquakes accurately located using the dense grid of seismographs in the Los Angeles region shows a north to NE trend that continues northward, beyond the surface trace of the west-trending Santa Monica fault. The earthquakes that are located to the north are deeper than the 9 September 2001 mainshock (M = 4.2; Hauksson et al. 2002). Geomorphic evidence for continuation of the blind Newport Inglewood fault zone farther to the NW is inconclusive but suggestive as the Sepulveda canyon crosses the Santa Monica Mountains in this area, although slightly to the west, consistent with the left-stepping en echelon pattern of the dextral strike-slip fault zone. The deep subsurface continuation of the Newport-Ingle wood fault zone may be responsible for segmentation of the west-trending faults within the San Fernando Valley. Bending fault termination The other end-member strike-slip fault termination style at convergence zone boundaries is represented by a broad curving of the strikeslip fault trace to merge with the cross-cutting structure. The strike-slip on the fault diminishes to zero at the termination, with the mass conservation requirement fulfilled by large-scale shortening to the SW of the NW-trending right-slip faults in folding and thrust or reverse faulting. Lateral escape of material by conjugate left-slip faulting may occur or vertical-axis rotation of local crustal blocks may accommodate the

diminishing right-slip (Wracher & Kamerling, pers. comm. 2003). For right-slip faults, the trace of the principal displacement zone bends to the left, suggesting local pure strain conditions where the conjugate faults bend away from the direction of principal shortening. Upper crustal material for this type of fault termination is unable to underthrust the cross-cutting structure and must be squeezed upward or outward and folded, resulting in crustal thickening near the intersection (Fig. 11). It is likely that midcrustal detachment occurs allowing the deeper crust and upper mantle lithosphere to underthrust the transverse structure (Yeats 1981; Wright 1991). Bending fault termination may be one symptom of this crustal delamination. The length of the curving fault section should be related to the depth to the detachment, with deeper detachment requiring greater distance for accommodation of greater material thickness in convergent structures. Triangular pull-apart basins are expected at the NE side of the fault termination as for the straight fault terminations, although the structural relief across the fault should be greater due to the more severe shortening to the SW of the fault termination. Polos Verdes fault zone. The Palos Verdes fault zone is part of a 300 km long major fault system within the Inner Borderland (Legg 1985; Jennings 1994). The Palos Verdes fault zone intersects the southern boundary of the WTR along a broadly curving trace across Santa Monica Bay (Fig. 9). Like the San Pedro Basin fault intersection with the WTR, the seafloor escarpment changes character from a

Fig. 11. Diagram showing model of shortening where the upper plate is detached and accommodates shortening by folding and thrust faulting, while the delaminated lower plate accommodates shortening by underthrusting and bulk thickening. This model is considered appropriate for bending strike-slip fault terminations. Local vertical-axis block rotations may also occur in the detached upper plate.

TERMINATION OF STRIKE-SLIP FAULTS

smooth gentle south slope to a steeper, more gullied slope at the projected intersection of the Palos Verdes fault. Large anticlinoria, and inferred thrust or reverse faults, occur on the southwestern side of the Palos Verdes fault zone for a distance of 60-75 km along strike, from San Pedro Bay to Santa Monica Bay (Nardin & Henyey 1978). These structures absorb the diminishing right-slip, allowing the gentle termination of the Palos Verdes fault zone at the Anacapa-Dume fault zone. A triangular pull-apart basin may occur in Santa Monica Bay, between the Shelf Projection Anticlinorium and the coast; it is nearly filled by sediments from the Los Angeles River, Ballona Creek branch and other local drainages of the WTR. The Palos Verdes fault zone cannot be easily traced with currently available data near the west-trending Anacapa-Dume or Santa Monica faults in Santa Monica Bay. Highresolution seismic imaging of the high-angle right-slip fault traces is difficult, although recently acquired profiles show lack of Holocene offset near Santa Monica Canyon (Fisher et al. 2001). To the SE of the Palos Verdes Hills, in San Pedro Bay, a 7 m dextral offset of a Late Holocene channel is mapped (McNeilan et al. 1996), so that about 3 m m a ^ 1 of right-slip must be accommodated as the fault dies out to the NW. Detachment of the Palos Verdes fault at shallow to mid-crustal levels is inferred (Davis et al. 1989; Shaw & Suppe 1996), although the fault is known to offset at shallow depth (less than 3 km) the top of the Catalina Schist basement, which probably represents a major Neogene detachment surface (Wright 1991; Crouch & Suppe 1993). The apparent disparity in Palos Verdes fault zone activity between San Pedro Bay and Santa Monica Bay may be related to significant shortening that has occurred in the block west of the fault. Four anticlinoria trend approximately 30 to 45 degrees west of the Palos Verdes fault trace; one of these is in San Pedro Bay, two are in Santa Monica Bay, and the last is the Palos Verdes Peninsula (Nardin & Henyey 1978). In addition, significant reverse displacement could have occurred along the Redondo Canyon fault, which diverges westward from the Palos Verdes fault zone north of the Palos Verdes Peninsula. One of the anticlinoria in Santa Monica Bay, the Shelf Projection Anticlinorium, is broken by many faults, including reverse faults, along the lower slope (Fisher et al. 2001). In contrast, the apron of sediments north of this anticlinorium, and north of Santa Monica Canyon, is little deformed except for two small anticlines that trend toward Point

77

Dume and are cored by reverse faults (Fisher etal 2001). The shortening features west of the Palos Verdes fault zone occur largely in a veneer of Neogene and Quaternary sediments above the Catalina Schist basement, a Mesozoic subduction complex (Wright 1991). The western edges of the four anticlinoria define the 60 km long escarpment that separates the continental shelf from Santa Monica and San Pedro basins. The westward protrusion of these anticlinoria suggest that they may be laterally extruded, in escape tectonics fashion (Walls et al. 1998). Surface sediments were forced to the west or NW buttressed against the high-angle, rightslip, Palos Verdes fault surface during NW transport of the Inner Borderland Schist basement under the WTR. The broad zone of microseismicity in this area (Hauksson & Saldivar 1989) may represent the more widely distributed convergent right-lateral shear on buried structures, possibly beneath an active detachment, due to the escape tectonics and accommodation of diminishing right-slip on the Palos Verdes fault zone. Whittier-Elsinore fault zone. The Whittier fault at the northern end of the Elsinore fault system curves broadly to the west from the southern N45°W trend to a N70°W trend where projected to merge with the Hollywood fault near downtown Los Angeles (Fig. 3; Jennings 1994). In detail, however, the Whittier fault turns to a more northerly trend along the East Montebello fault (Figs 9 & 10; Wright 1991) and right-slip diminishes almost completely before reaching the southern boundary of the Transverse Ranges at the Hollywood fault. Unlike the Palos Verdes fault, however, the Whittier fault bends abruptly south of Corona, where the Chino fault splits off and follows a NW trend subparallel to the Elsinore fault. Therefore, rather than forming a smooth bending fault termination, it is possible that the Whittier fault is the surface expression of a major restraining bend along the Elsinore fault. The elevated Puente Hills and Chino Hills between the Whittier and Chino fault zones may be considered pop-up structures in the major restraining bend (McClay & Bonora 2001), whereas the bending fault termination model predicts that uplift from fold-and-thrust related shortening would be located to the SW of the major right-slip fault. Several west-trending folds and blind thrust faults are located to the SW of the Whittier fault, including the Puente Hills thrust and the Elysian Park fault (Wright 1991; Shaw & Shearer 2000; Oskin et al. 2000; Shaw et al. 2002). These may be associated with the

78

M. R. LEGGETAL

termination of the Whittier fault zone at the WTR boundary. Thus, the Whittier fault termination appears to be a combination of strike-slip fault termination styles. The NW projection of the Whittier-Elsinore fault zone, including the Chino-Central Avenue faults, would follow the Sierra Madre, San Gabriel and Verdugo fault zones in the WTR (Fig. 10). Although this alignment is suggestive of deep crustal structural control, it is uncertain at this time what, if any, relationship exists between the NW-trending Peninsular Ranges faults and the NW-trending Transverse Ranges faults in this area. Vertical coupling and decoupling along the southern Western Transverse Ranges boundary Palaeomagnetic data show that the WTR have rotated as an elongate coherent block about a vertical axis from 75 to 90 degrees in a clockwise sense (Kamerling & Luyendyk 1979, 1985; Luyendyk et al 1985; Luyendyk, 1991) since about 18 Ma. Two contrasting models have been used to explain the driving mechanism of this rotation (Fig. 4): (1) edge-driven rotation wherein the edges of the adjacent non-rotating blocks force the rotation of the intervening WTR block to rotate clockwise in the regional shear couple along the transform system (Luyendyk et al. 1980); and (2) basal shear tractions imparted by microplate capture during Neogene transtension along the evolving Pacific-North America transform margin (Nicholson et al 1994). It was recognized that Neogene transtension was necessary to facilitate the large-scale rotation of the elongate WTR block without westward thrusting over the adjacent Pacific plate (Legg 1991; Luyendyk 1991), but post-Miocene NE-directed shortening may have stopped the block rotation. Alternatively, once the rotation passed the point of maximum S W projection of the WTR block, when orthogonal to the San Andreas fault system, continued clockwise rotation would be compatible with NE-directed shortening across the plate boundary (Luyendyk 1991). The termination of NW-trending strike-slip faults in the non-rotated California Continental Borderland and adjacent coastal Peninsular Ranges provides important clues as to the mechanism of the block rotations and whether these rotations continue in Quaternary time. Edge-driven block rotation For rigid block rotation, the slip rates on the nonrotated, right-slip faults driving the clockwise

vertical-axis rotation of the WTR may be predicted from the angular rate of rotation and distance of each right-slip fault from the pivot point of the block rotation (Fig. 4a; Luyendyk et al. 1980). The net NW motion of the nonrotated blocks, relative to stable North America, increases systematically westward from the pivot point, and must include the net NW motion of that pivot point, which is unknown at present. The relative motion across each block boundary, as represented by the slip rate on each NW-trending right-slip fault, is only a function of the block width normal to the fault. Wider blocks, like the Santa Monica Basin block between the San Pedro Basin and Santa Cruz-Catalina Ridge fault zones, require higher slip rates to accommodate the clockwise WTR rotation than do narrow blocks, like the Santa Monica Bay block between the Palos Verdes and San Pedro Basin fault zones. Therefore, in the edge-driven rigid block rotation model, the slip rate on the Santa Cruz-Catalina Ridge fault zone should be greater than that of the San Pedro Basin fault zone and of the Palos Verdes fault zone. Unfortunately, at present, only the slip rate on the Palos Yerdes fault zone has been quantified with sufficient accuracy to compare slip rates to predictions. Furthermore, this rate varies from 3 mm a~ l in the Los Angeles harbour region (McNeilan et al. 1996) and diminishes to near zero in Santa Monica Bay (Fisher et al. 2001). Geologically, the San Clemente fault system, which includes the Santa Cruz-Catalina Ridge fault zone, appears very well defined, long and continuous, and therefore may have a significant slip rate (about 4-7 mm a"1, Legg 1985). Seismicity is also greatest along the San Clemente fault system including the 1981 Santa Barbara Island earthquake (M=6.0; Corbett 1984; Bent & Helmberger 1991). Yet, the geodetic data find relatively little NW right-shear between Santa Catalina Island and San Nicolas Island (Bennett et al. 1996), which should constrain the rate of slip on the San Clemente fault system. Consequently, at this time the validity of the edge-driven model cannot be conclusively demonstrated with quantitative data. In addition to the predictable slip rates for nonrotated fault blocks, the edge-driven model predicts the formation of triangular-shaped pullapart basins on the NE side of the right-slip fault terminations (Luyendyk et al. 1980). As with the relative magnitude of the slip rates, the size of these triangular basins is also proportional to the width of the non-rotated block. Qualitatively, this model prediction appears to work for largescale block rotations, with Santa Monica Basin being one of the largest of the borderland basins,

TERMINATION OF STRIKE-SLIP FAULTS

followed by the Los Angeles Basin - one of the deepest borderland basins. On a smaller scale, local basins occur at the NE side of the fault terminations for the Santa Cruz-Catalina Ridge fault zone and for the Newport-Inglewood fault zone as described above. In Santa Monica Bay, east of the San Pedro Basin fault zone, a basin exists, but appears to cross the Palos Verdes fault zone, which has a bending strike-slip fault termination, and therefore may be inappropriate for the rigid edge-driven model. In contrast to the triangular pull-apart basins, well-defined folding and thrust faulting occur on the SW side of the right-slip fault terminations, e.g. south of Santa Cruz Island and south of the Santa Monica fault zone. For the edge-driven model, deformation of the edges of the NW-trending blocks should occur where the right-slip exceeds that predicted by the rate of clockwise rotation and local NW translation of the WTR at the fault intersection. But this observation is more consistent with the basal sheardriven model in that some part of the borderland crust underthrusts the WTR and is available to drive rotations through the shear couple induced by the 'blind' strike-slip faulting or shear zone (Figs 4b & 5). Tsutsumi et al (2001) find that the amount of right-slip accommodated by folding west of the Inglewood fault is only about one-third of the amount measured in the Inglewood oil field; the remainder may be underthrust and provide clockwise basal torque under the Santa Monica Mountains. At the greater depths expected for the underthrust Inner Borderland block, a basal shear zone rather than a discrete narrow fault zone is more likely in the lower plate beneath the WTR. Presumably, the underthrust block extends below the brittle-ductile transition in the region and would appear relatively aseismic. Basal shear-driven block rotation According to the microplate capture model of Nicholson et al (1994), the WTR is detached above a mid-crustal decollement and rotates clockwise about a vertical axis due to the constraint of the eastern end of the block by irregularity in the transform boundary (Fig. 4b). That constraint also defines the pivot point of the WTR vertical-axis rotation. For complete microplate capture, i.e. the entire Monterey microplate became rigidly attached to the NW-moving Pacific plate with no internal shear, the rotation of the WTR requires the constraint of the eastern end of the block. In this case, basal traction by the captured microplate would be uniform from west to east with no induced clockwise torque. Instead, the overlying

79

clockwise block rotation would induce a laterally varying drag that is greatest to the east where the NW component of WTR block motion is a minimum. If, instead of complete capture, the microplate breaks up into transform parallel slivers, bounded by the NW-trending right-slip faults of the non-rotated Inner Borderland, the basal tractions due to the microplate capture include a dextral shear component that induces a clockwise torque on the overlying WTR (Figs 4b & 5). In this case, the constraint at the eastern end of the WTR block is not required, and the pivot point of the vertical-axis rotation may be located elsewhere. Furthermore, as the rotation proceeds to the point where the WTR becomes nearly orthogonal to the NW translation of the Pacific plate, the obstruction to smooth transform motion may cause the WTR to break up or rotate differently. The rotation may proceed in separate pieces of the WTR block, as proposed by Hornafius (1985), or at a different rate for the entire WTR block. The tendency to break up the WTR would be enhanced by the westward migration of the Sierra Nevada block due to Basin and Range extension and growth of the major restraining bend of the San Andreas fault (Bird & Rosenstock 1984). The increased resistance to NW translation of the southern California blocks west of the San Andreas fault past the WTR would increase the vertical coupling across high-angle fault systems along the southern boundary of the WTR. This resistance could lead to increased fault locking between large thrust or oblique-reverse earthquakes on faults such as the Sierra Madre, Cucamonga, Raymond, Hollywood, Santa Monica, AnacapaDume, Channel Islands and Santa Cruz Island fault zones. Blocks in the upper plate, above the regional detachment system that underthrusts the WTR, would push the WTR as in the edgedriven model, or require some form of 'escape tectonics' as proposed by Walls et al. (1998) to extrude laterally the excess material entering the convergence zone. Left-slip on west-trending and NE-trending high-angle faults is considered the most likely mechanism to allow this lateral extrusion, resulting in westward elongation of the WTR during the vertical-axis clockwise rotation. Observations and conclusions Two end-member styles of strike-slip fault termination at the southern boundary of the WTR convergence zone are apparent in southern California. NW-trending right-slip faults that terminate with a generally straight surface trace at the boundary are considered to be continuous with lower plate faults, beneath a mid-crustal detachment, that facilitate the clockwise

80

M. R. LEGGETAL.

vertical-axis rotation of the WTR. These highangle strike-slip faults, therefore, offset the lower plate beneath the regional detachment. Geophysical data provide compelling evidence that these blind strike-slip faults do exist for some distance, at least several tens of kilometres beneath the WTR. Owing to the depth of the detachment beneath the WTR, these blind right-slip faults likely exist as ductile shear zones, but still provide a clockwise torque, by basal shear at the detachment beneath the WTR. This clockwise torque provides one mechanism to maintain clockwise vertical-axis rotation of the WTR. The relative efficiency of this torque depends upon the strength of the coupling across the detachment. A more quantitative measure of that coupling may be estimated by comparing the predicted right-slip fault slip rates from the edge-driven model versus the measured slip rates, with the difference being required for underthrusting and fold development in the hanging wall of the Channel Islands thrust. Right-slip faults that terminate by broadly curving westward to merge into the WTR boundary fault system are considered to be confined to the upper crust above a regional detachment. Below the detachment, the lower plate delaminates and may underthrust the WTR. The wavelength of the fault curvature is most likely related to the depth of detachment, but more thorough modelling is required to quantify this relationship. These bending faults die out, with diminishing right-slip being accommodated by folding and thrust or reverse-faulting along the SW side of the principal displacement zone. Local vertical-axis block rotations may occur to accommodate the slip transfer, and the westward bending is consistent with pure shear where conjugate strike-slip faults rotate away from the direction of principal shortening. Deeper dextral shear zones, in the lower plate beneath the regional detachment that underthrusts the WTR, may not necessarily lie directly beneath the upper plate right-slip fault zones. Consequently, one cannot directly infer the geometry of the deep crustal fault or shear zones based upon the detached upper plate fault zone. Broad patterns of seismicity in the coastal and near-offshore Los Angeles region are consistent with this detached character, with focal mechanisms varying from strike-slip to thrust faulting and few oblique-slip mechanisms (Hauksson 1990; Hauksson & Saldivar 1989). Indeed, the focal mechanisms of earthquakes in this area provide convincing evidence of strain partitioning in the upper crust (Nicholson & Crouch 1989). The high-angle NewportIngle wood fault zone accommodates right-slip,

and the broad diffuse zone of seismicity SW of the Palos Verdes fault zone and NE of the San Pedro Basin fault zone accommodates NEdirected shortening (Nicholson & Crouch 1989; Hauksson 1990). This shortening is considered to result from the dissipation of right-slip on the Palos Verdes fault zone in the detached, upper crustal region. Deeper earthquakes, below the regional detachment, may also accommodate NE-directed shortening, but more accurate hypocentral locations are required to determine where the regional detachment lies relative to the seismicity and whether any distinctive separation between upper plate strain partitioning and lower plate bulk shortening may be recognizable. This research was supported by the Southern California Earthquake Center. SCEC is funded by NSF Cooperative Agreement EAR-8920136 and USGS Cooperative Agreements 14-08-0001-A0899 and 1434-HQ-97AG01718. The SCEC contribution number for this paper is 678.

References BARROWS, A.G. 1974. A Review of the Geology and Earthquake History of the Newport-Inglewood Structural Zone, Southern California. California Division of Mines & Geology Special Report 114, 115pp. BENNETT, R.A., RODI, W. & REILINGER, R.E. 1996. Global Positioning System constraints on fault slip rates in southern California and northern Baja, Mexico. Journal of Geophysical Research, 101, 21 943-21 960. BENT, A.L. & HELMBERGER, D.V. 1991. Seismic characteristics of earthquakes along the offshore extension of the Western Transverse Ranges, California. Seismological Society of America Bulletin, 81, 399-422. BIRD, P. & ROSENSTOCK, R.W. 1984. Kinematics of present crust and mantle flow in southern California. Geological Society of America Bulletin, 95, 946-957. BOHANNON, R.G. and GEIST, E. 1998. Upper crustal structure and Neogene tectonic development of the California Continental Borderland. Geological Society of America Bulletin, 110, 779-800. BURDICK, D.J. & RICHMOND, W.C. 1982. A Summary of Geologic Hazards for Proposed OCS Oil and Gas Lease Sale 68, Southern California. US Geological Survey Open-File Report 82-33. CHRISTIE-BLICK, N. & BIDDLE, K.T. 1985. Deformation and basin formation along strike-slip faults. In: BIDDLE, K.T. & CHRISTIE-BLICK, N. (eds) StrikeSlip Deformation, Basin Formation, and Sedimentation. Society of Economic Paleontologists and Mineralogists Special Publication 37, 1-34. CORBETT, EJ. 1984. Seismicity and crustal structure studies of southern California - Tectonic implications from improved earthquake locations. PhD thesis, California Institute of Technology, Pasadena.

TERMINATION OF STRIKE-SLIP FAULTS CROUCH, J.K. & SUPPE, J. 1993. Late Cenozoic tectonic evolution of the Los Angeles basin and inner California borderland: a model for core complex-like crustal extension. Geological Society of America Bulletin, 105, 1415-1434. CROWELL, J.C. 1974. Origin of late Cenozoic basins in southern California. In: DICKINSON, W.R. (ed.) Tectonics and Sedimentation. Society of Economic Paleontologists and Mineralogists Special Publication 22, 190-204. DARTNELL, P. & GARDNER, J.V. 1999. Seafloor images and data from multibeam surveys in San Francisco Bay, southern California, Hawaii, the Gulf of Mexico, and Lake Tahoe, CaliforniaNevada. US Geological Survey Digital Data Series DDS-55, version 1 (CD-ROM). DAVIS, T.L., NAMSON, J. & YERKES, R.F. 1989. A cross-section of the Los Angeles area: seismically active fold-and-thrust belt, the 1987 Whittier Narrows earthquake, and earthquake hazard. Journal of Geophysical Research, 94, 9644-9664. FISHER, M.A., NORMARK, W.R., BOHANNON, R.G., SLITER, R.W. & CALVERT, A.J. 2001. Geology of the continental margin beneath Santa Monica Bay from small-airgun seismic-reflection data. Transactions, American Geophysical Union, 82, F804. FRANCIS, R.D., SIGURDSON, D.R., LEGG, M.R., GRANNELL, R.B. & AMBOS, E.L. 1999. Student participation in an offshore seismic reflection study of the Palos Verdes fault, California Continental Borderland. Journal of Geo-Science Education, 47, 22-30. FREUND, R. 1971. The Hope Fault - A Strike-Slip Fault in New Zealand. New Zealand Geological Society Bulletin, 86, 49 pp. HARDING, T.P., VIERBUCHEN, R.C. & CHRISTIEBLICK, N.C. 1985. Structural styles, plate-tectonic settings, and hydrocarbon traps of divergent (transtensional) wrench faults. In: BIDDLE, K.T. & CHRISTIE-BLICK, N. (eds) Strike-Slip Deformation, Basin Formation, and Sedimentation. Society of Economic Paleontologists and Mineralogists Special Publication 37, 51-77. HAUKSSON, E. 1990. Earthquakes, faulting, and stress in the Los Angeles basin. Journal of Geophysical Researches, 15365-15394. HAUKSSON, E. & SALDIVAR, G.V. 1989. Seismicity and active compressional tectonics in Santa Monica Bay, southern California. Journal of Geophysical Research, 94, 9591-9606. HAUKSSON, E., JONES, J., PERRY, S. & HUTTON, K. 2002. Emerging from the stress shadow of the 1992 Mw Landers southern California earthquake? A preliminary assessment. Seismological Research Letters, 73, 33-38. HORNAFIUS, J.S. 1985. Neogene tectonic rotation of the Santa Ynez Range, Western Transverse Ranges, California, suggested by paleomagnetic investigation of the Monterey formation. Journal of Geophysical Research, 90, 1250312522. HORNAFIUS, J.S., LUYENDYK, B.P., TERRES, R.A. & KAMERLING, M.J. 1986. Timing and extent of Neogene tectonic rotation in the Western Transverse

81

Ranges, California. Geological Society of America Bulletin, 97, 1476-1487. JENNINGS, C.W. 1994. Fault activity map of California and adjacent areas. California Division of Mines and Geology, Map No. 6, scale 1:750000. JUNGER, A. 1979. Maps and seismic profiles showing geologic structure of the northern Channel Islands platform, California Continental Borderland. US Geological Survey, Miscellaneous Field Studies, Map MF-991, scale 1:250000. JUNGER, A. & WAGNER, H.D. 1977. Geology of the Santa Monica and San Pedro Basins, California Continental Borderland. US Geological Survey, Miscellaneous Field Studies, Map MF-820, scale 1:250000. KAMERLING, M. & LUYENDYK, B.P. 1979. Tectonic rotations of the Santa Monica Mountains region, Western Transverse Ranges, California, suggested by paleomagnetic vectors. Geological Society of America Bulletin, 90, 331-337. KAMERLING, M. & LUYENDYK, B.P. 1985. Paleomagnetism and Neogene tectonics of the northern Channel Islands, California. Journal of Geophysical Research, 90, 12485-12502. LANGENHEIM, V.E., HALVORSON, P.P., CASTELLANOS, E.L. & JACHENS, R.C. 1993. Aeromagnetic map of the Southern California Borderland east of the Patton Escarpment. US Geological Survey OpenFile Report 93-520, map scale 1:500000. LEGG, M.R. 1985. Geologic structure and tectonics of the inner continental borderland offshore northern Baja California, Mexico. PhD thesis, University of California, Santa Barbara. LEGG, M.R. 1991. Developments in understanding the tectonic evolution of the California Continental Borderland. In: OSBORNE, R.H. (ed.) From Shoreline to Abyss: Contributions in Marine Geology in Honor of Francis Parker Shepard. SEPM Special Publication 46, Tulsa, OK, 291-312. LONSDALE, P. 1991. Structural patterns of the Pacific floor offshore of Peninsular California. In: DAUPHIN, P. & NESS, G. (eds) The Gulf and Peninsular province of the Californias. American Association of Petroleum Geologists Memoir 47, Tulsa, OK, 87-125. LUYENDYK, B.P. 1991. A model for Neogene crustal rotations, transtension, and transpression in southern California. Geological Society of America Bulletin, 103, 1528-1536. LUYENDYK, B.P., KAMERLING, M.J. & TERRES, R.R. 1980. Geometric model for Neogene crustal rotations in southern California. Geological Society of America Bulletin, 91, 211-217. LUYENDYK, B.P., KAMERLING, M.J., TERRES, R.R. & HORNAFIUS, J.S. 1985. Simple shear of southern California during Neogene time suggested by paleomagnetic declinations. Journal of Geophysical Research, 90, 12454-12466. McCLAY, K. & BONORA, M. 2001. Analog models of restraining stepovers in strike-slip fault systems. American Association of Petroleum Geologists Bulletin, 85, 233-260.

82

M. R. LEGGETAL.

MCNEILAN, T.W., ROCKWELL, T.K. & RESNICK, G.S. 1996. Style and rate of Holocene slip, Palos Verdes fault, southern California. Journal of Geophysical Research, 101, 8317-8334. NARDIN, T.R. & HENYEY, T.L. 1978. PliocenePleistocene diastrophism of the Santa Monica and San Pedro shelves, California Continental Borderland. American Association of Petroleum Geologists Bulletin, 62, 247-272. NICHOLSON, C. & CROUCH, J.K. 1989. Neotectonic structures along the central and southern California margin. Seismological Research Letters, 60, 23-24. NICHOLSON, C., SORLIEN, C.C., ATWATER, T., CROWELL, J.C. & LUYENDYK, B.P. 1994. Microplate capture, rotation of the Western Transverse Ranges, and initiation of the San Andreas transform as a low-angle fault system. Geology, 22, 491—495. OSKIN, M., SIEH, K., etal 2000. Active parasitic folds on the Elysian Park anticline: implications for seismic hazard in central Los Angeles, California. Geological Society of America Bulletin, 112, 693-707. RICHMOND, W.C., CUMMINGS, L.J., HAMLIN, S. & NAGATY, M.E. 1981. Geologic Hazards and Constraints in the Area of Proposed OCS Oil and Gas Lease Sale 48, Southern California. US Geological Survey Open-File Report 81-307. SHAW, J. & SHEARER, P. 2000. An elusive blind-thrust fault beneath metropolitan Los Angeles. Science, 283, 1516-1518. SHAW, J.H. & SUPPE, J. 1996. Earthquake hazards of active blind-thrust faults under the central Los Angeles basin, California. Journal of Geophysical Research, 101, 8623-8642. SHAW, J.H., PLESCH, A., DOLAN, J.F., PRATT, T.L., & FIORE, P. 2002. Puente Hills blind-thrust system, Los Angeles, California. Bulletin of the Seismological Society of America, 92, 2946-2960.

SHOR, G.G., JR. & RAITT, R.W. 1958. Seismic studies in the southern California borderland. Congreso Geologico Internacional, XXa Seccion, 1956, Mexico, Seccion IV, Segundo Tomo, 243-259. SORLIEN, C.C., KAMERLING, M.J. & SEEBER, L. 2001. The Dume fault, northern Santa Monica Bay, California. Transactions, American Geophysical Union, 82, F802. TEN BRINK, U.S., ZHANG, J., BROCHER, T.M., OKAYA, D.A., KLITGORD, K.D. & Fuis, G.S. 2000. Geophysical evidence for the evolution of the California Inner Continental Borderland as a metamorphic core complex. Journal of Geophysical Research, 105, 5835-5857. TSUTSUMI, H., YEATS, R.S. & HUFTILE, GJ. 2001. Late Cenozoic tectonics of the northern Los Angeles fault system, California. Geological Society of America Bulletin, 113, 454-468. VEDDER, J.G., GREENE, H.G., CLARKE, S.H. & KENNEDY, M.P. 1986. Geologic map of the midsouthern California continental margin. California Division of Mines & Geology, California Continental Margin Geologic Map Series, Area 2 of 7, sheet 1 of 4, scale 1:250000. WALLS, C., ROCKWELL, T.K. et al 1998. Escape from L.A.: extrusion tectonics in southern California and implications for seismic risk. Nature, 394, 356-360. WGCEP (Working Group on California Earthquake Probabilities) 1995. Seismic hazards in southern California: probable earthquakes, 1994 to 2024. Bulletin of the Seismological Society of America, 85, 379-439. WRIGHT, T.L. 1991. Structural geology and tectonic evolution of the Los Angeles basin. In: BIDDLE, K.T. (ed.) Active Margin Basins. American Association of Petroleum Geologists Memoir 52, 13-134. YEATS, R.S. 1981. Quaternary flake tectonics of the California Transverse Ranges. Geology, 9, 16-20.

Vertical-axis rotation of rigid crustal blocks driven by mantle flow SCOTT GIORGIS1, MICHELLE MARKLEY2 & BASIL TIKOFF1 1 Department of Geology and Geophysics, University of Wisconsin, Madison, WI 53706, USA (e-mail: [email protected]) 2Department of Earth and Environment, Mount Holyoke College, South Hadley, MA 01075, USA Abstract: Vertical-axis rotation of rigid crustal blocks occurs in a variety of obliquely convergent and divergent plate boundaries. We quantify the rotation of these blocks using models of transpressional and transtensional kinematics, and corroborate our results using physical models where rigid blocks rotate in response to flow of a ductile substrate. Consequently, one can explicitly demonstrate a relationship between the amount of rotation of a rigid crustal block and strain recorded in ductile substrate. This strain should be reflected directly by the orientation of rock fabrics, such as those measured by shear-wave splitting in the in situ upper mantle.s We apply this approach to southern California and New Zealand by using previously documented palaeomagnetic rotations and plate motion vectors, and calculate the strain recorded by the material below rigid blocks. These strain calculations are compared to shear-wave splitting data, which record upper mantle fabric, from the same region. Our model results suggest that similar deformation is recorded by the upper crust and lithospheric mantle. A bottom-driven flow, in which mantle deformation drives upper crustal rotations, is most consistent with these observations.

The nature of the boundary conditions controlling deformation in the crust remains a topic of debate in plate tectonic theory. It is commonly assumed that horizontal plate motion (side-driven boundary conditions) governs plate interaction. An alternative school of thought emphasizes how upper crustal deformation may be due to flow in the mantle (bottom-driven systems, e.g. McKenzie & Jackson 1983; Molnar 1992; Teyssier & Tikoff 1998). One recent adaptation of this debate (clutch tectonics) calls on partial attachment zones to transfer mantle flow into upper crustal deformation or upper crustal deformation into mantle flow (Teyssier et al 2002; Tikoff et al 2002). There are four end-member cases for mantle crust interaction (Fig. 1): (1) perfect detachment (no interaction between upper crust and mantle); (2) a top-driven system (upper crustal deformation drives mantle flow); (3) perfect attachment (sidedriven systems); and (4) a bottom-driven system (mantle flow drives upper crustal deformation; Tikoff et al 2002). Figure 1 shows these endmember cases in the context of a rigid, cylindrical block rotating above a flowing mantle, both of which are deforming in simple shear. For each of the above scenarios, the different endmembers make predictions of the geometry of finite strain in the mantle for a given amount of

vertical-axis rotation of the upper crustal block (Teyssier et al 2002). These predictions offer a way to test whether bottom-driven or side-driven boundary conditions govern deformation. The simplicity of the predictions makes them amenable to analysis with numerical and physical modelling. Analytical solutions to the rotation of rigid elliptical bodies over a viscously deforming medium (e.g. Ghosh & Ramberg 1976; Tikoff & Teyssier 1994) relate the amount of rotation of the rigid ellipse to the magnitude and orientation of finite strain in the deforming substrate. Physical models of these systems in simple shear and pure shear environments indicate that the equations accurately describe this relationship (e.g. Ghosh & Ramberg 1976; Arbaret et al 2000). Crustal plate boundaries, however, are more accurately described by the kinematic models of transpression and transtension than pure shear or simple shear (Sanderson & Marchini 1984; Dewey et al 1998). We approach this problem by first testing the analytical model (Jeffery 1922; Ghosh & Ramberg 1976) against physical models of rigid object rotation in transpression and transtension. We then apply these equations to natural systems, specifically the San Andreas fault system in California and the Alpine fault

From: GROCOTT, J., MCCAFFREY, K. J. W., TAYLOR, G. & TIKOFF, B. (eds) 2004. Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 83-100. 0305-8719/047$ 15 © The Geological Society of London 2004.

Fig. 1. Schematic relationship between a rigid, cylindrical crustal block that has experienced a 90° vertical-axis rotation and an underlying ductile layer, (a) Perfect detachment - there is no interaction between the upper rotating block and the lower ductile layer, (b) Top-driven system - the rigid block is 'grabbed from the sides' and rotating, imposing deformation on the lower ductile layer, (c) Perfect attachment (side-driven system) - both the rigid block and the ductile layer are 'grabbed from the sides' and subject to the same deformation boundary conditions, (d) Bottom-driven system - flow in the ductile layer dictates the rotation rate of the upper crustal block. Below each diagram is a stereogram depicting the orientation of finite strain in the ductile layer and the predicted orientation and relative magnitude of the fast direction of shear-wave splitting. Orientations calculated based on simple shear boundary conditions. See text for details. Figure modified from Teyssier el al. (2002).

CRUSTAL BLOCK ROTATION BY MANTLE FLOW

85

system in New Zealand. Both of these plate boundaries have rigid crustal blocks that have experienced vertical-axis rotation, the magnitude of which was constrained using palaeomagnetism. We then calculate the predicted orientation of finite strain in the lithospheric mantle based on modern plate motions and the amount of block rotation. Shear-wave splitting appears to give information about the orientation and magnitude of finite strain in the mantle (e.g. Silver 1996; Savage 1999). Therefore we test our modelled predictions of fabric orientation in the mantle against the shear-wave splitting datasets to distinguish between perfect detachment, topdriven, perfect attachment (side-driven) and bottom-driven systems (Fig. 1). Physical models Rigid rotation: boundary conditions and background The purpose of these physical modelling experiments is to test whether or not the Jeffery (1922) model for rotation of rigid objects in a ductile flow predicts the rotation rates of rigid objects on top of a ductile substrate. In these experiments, we explore a variety of displacement-controlled boundary conditions for the substrate: convergence, oblique convergence, transcurrence, oblique divergence and divergence (Fig. 2). Although we refer to these deformations as transcurrence and convergence/divergence in order to associate them with regional tectonic regimes, we use simple shear, pure shear, transtension and transpression to model the kinematics of the ductile substrate. The rotation of material lines during deformations described by these boundary conditions is well understood (e.g. Fossen et al 1994). In all these deformations, lines that are initially parallel to either the z-axis (vertical) or the jc-axis remain parallel to that axis during deformation. In transpression and convergent pure shear, initially inclined material lines rotate towards vertical, and lines in the horizontal (xy) plane are metastable. In other words, initially horizontal lines generally remain in the horizontal plane even though it is not a stable orientation. In simple shear, transtension and divergent pure shear, initially inclined material lines rotate towards the xy (horizontal) plane, and this plane is a stable orientation (i.e. initially horizontal lines remain horizontal). In particular, we focus on the rotation of material lines in the horizontal plane (Fig. 3). During convergent pure shear (Fig. 3a) and transpression (Fig. 3b), two orientations of lines do

Fig. 2. The kinematic regimes we explore combine wrenching with convergence and divergence. The coordinate system we use is as follows: a is the obliquity of convergence/divergence with respect to the rigid boundaries, x is the horizontal direction of wrenching, y is also horizontal and perpendicular to jc, and z is vertical, (a) a— — 90° for convergence of the rigid boundaries (the darker boxes). Convergence causes vertical thickening and homogeneous pure shear in the deforming zone (the lighter grey box). Flow is confined to the yz plane, (b) 0° > a > —90° in dextral oblique convergence, which causes homogeneous transpression in the deforming zone. The shear plane parallels the xz plane, (c) a = 0° for dextral wrenching (+180° for sinistral wrenching), which causes homogeneous simple shear in the deforming zone. The shear plane parallels the xz plane, (d) 0° < a < 90° for dextral oblique divergence, which causes homogeneous transtension in the deforming zone, (e) a = 90° for divergence. Divergence causes vertical thinning and homogeneous pure shear in the deforming zone. Flow is confined to the yz plane.

86

S. GIORGISCTAL.

Fig. 3. Rotation of material lines in the horizontal (xy) plane during deformation, based on work by Ramberg (1975), Fossen et al (1994) and Passchier (1997). (a) In convergent pure shear, the repulsor flow apophysis parallels the y-axis, and the attractor flow apophysis parallels the ;c-axis. Half the material lines rotate clockwise, and half rotate counterclockwise, (b) In transpression, the repulsor parallels the direction of convergence, and the attractor parallels the x-axis. Most, but not all, lines rotate clockwise, (c) In simple shear, the z-axis acts as both an attractor and a repulsor, and all lines rotate clockwise, (d) In transtension, the repulsor parallels the z-axis, and the attractor parallels the direction of divergence. Most, but not all, lines rotate clockwise, (e) In divergent pure shear, the repulsor parallels the Jt-axis, and the attractor parallels the y-axis. Half the material lines rotate clockwise, and half rotate counterclockwise.

not rotate: those that are initially parallel to the xaxis and those that are initially parallel to the direction of convergence (at 90° in Fig. 3a and 135° in Fig. 3b). Other material lines rotate towards the jc-axis and away from the direction of convergence. The jc-axis is an 'attractor' flow apophysis, and the direction of convergence is a 'repulsor' flow apophysis (Ramberg 1975; Fossen et al. 1994; Passchier 1997). During simple shear (Fig. 3c), lines that are initially parallel to the *-axis (the shear plane) do not rotate, and all other horizontal lines rotate in the same direction (clockwise in the example shown here) towards this orientation. The ;t-axis is therefore both an attractor and a repulsor. During oblique divergence (Fig. 3d) and divergent pure shear (Fig. 3e), two orientations of lines again do not rotate: those that are initially parallel to the *-axis and those that are initially parallel to the direction of divergence (at 45° in Fig. 3d and 90° in Fig. 3e). In this case the direction of divergence acts as an attractor and the Jt-axis acts as a repulsor. In all deformations except for simple shear, the sense of rotation of a material line (clockwise or counterclockwise) depends on the initial orientation of the line. Rigid objects embedded in viscous, deforming media are also sensitive to fabric attractors and repulsors (Jeffery 1922). Rotation rates for rigid ellipses are well established in simple shear and pure shear (Ghosh & Ramberg 1976; Willis

1977), and through sparse work on other kinematic regimes such as transpression and transtension (Freeman 1985; Jezek et al 1994, 1996). The mathematics used in this study to calculate the orientations of rigid clasts in transpression (and transtension) are described in Giorgis & Tikoff (2004). The behaviour of rigid clasts with low aspect ratios (< 10:1) differs from that of material lines in that there are no orientations in which rigid ellipses do not rotate (Ghosh & Ramberg 1976). Figure 4 shows two examples in transpression for an ellipse whose aspect ratio is 1.75:1. Figures 2b and 3b are useful for comparison because they explore the same boundary condition: transpression related to oblique convergence with a = — 4 5 ° . For material lines, the sense of rotation depends on the initial orientation (Fig. 3b). For example, a material line initially oriented at 20° from the positive x-axis rotates clockwise, but one at 160° rotates counterclockwise. But for the lowaspect-ratio rigid ellipses that we consider in dextral transpression, the sense of rotation is always clockwise (Fig. 4b). This observation is not true for all low-aspect-ratio rigid objects in transpression. For example, ellipses of this aspect ratio show both senses of rotation during transpression related to higher angles of convergence (a closer to —90°). The aspect ratio of a rigid object also exerts a first-order control on its rotation rate and stable positions (Jeffery 1922; Ghosh & Ramberg

CRUSTAL BLOCK ROTATION BY MANTLE FLOW

Fig. 4. Rotation of a rigid ellipse with an aspect ratio 1.75:1, in the horizontal plane during transpression, based on work by Ghosh & Ramberg (1976). (a) In highly oblique transpression (a — — 15°), ellipses of any initial orientation rotate clockwise. An ellipse oriented parallel to the fabric attractor (0° or 180°) or repulsor (165°) rotates more slowly than others, (b) In less oblique transpression (a = -45°), ellipses of any initial orientation rotate clockwise. An ellipse oriented parallel to the fabric attractor (0° or 180°) or repulsor (135°) rotates more slowly than others.

1976). Equant rigid objects rotate at a rate dependent only on the vorticity and strain rate of the viscously deforming medium. The rotation rates of low-aspect-ratio rigid objects (<10:1) also depend on initial orientation (Fig. 4). Equant and low-aspect-ratio objects do not rotate into a stable position, unlike material lines, which rotate into parallelism with the fabric attractor flow apophysis (Ghosh & Ramberg 1976). The physical experiments of Ghosh & Ramberg (1976) showed that objects with high aspect ratios (>10:1) rotate similarly to material lines. For example, higher-aspect-ratio rigid objects than the one considered in Figure 3b show both senses of rotation during transpression related to the same boundary conditions (a = —45°). Experimental apparatus and design Our experimental apparatus (Fig. 5) is described in Venkat-Ramani & Tikoff (2002). These experiments use a rubber sheet and silicone gel to distribute simple shear, pure shear, transpression or transtension homogeneously in a basal rubber sheet. The rubber sheet stretches to approximately twice its original width without buckling or breaking. In these experiments, we covered the rubber sheet with an even thickness of silicone (RD-20; Rhone Poulenc), which behaves in a Newtonian fashion (viscosity = 1 x 104 Pa s)

87

and sticks to the underlying sheet. The initial thickness of this layer of silicone was approximately 2 cm. We embedded two different kinds of rigid objects into the surface of the silicone and gathered a variety of data from video stills in order to document the experiments. First, we performed a series of experiments involving matchsticks, which have an aspect ratio of ^25:1 (Fig. 6). On each video still, we measured matchstick orientation (/3) and the position of the moving plate. Comparing the initial width (w) of the deforming zone in the y direction, the change in position (dy) of the moving plate in the y direction and the change in position (dx) of the moving plate in the x direction, we calculated a dimensionless measure of plate displacement:

From <2, we calculated the axial ratio of the strain ellipse in the xy plane (Rf). The deformations we explored with this first suite of experiments included convergence (a = — 90°), oblique convergence (a = — 4 5 ° and —15°), wrenching (a = 180°), oblique divergence (a = 165° and 135°) and divergence (a = 90°). A second series of experiments used rigid Plexiglas ellipses with a thickness of about 1 cm and aspect ratios of 1.5:1, 1.75:1 and 2:1 (Fig. 7). In these experiments, we drew a circle on the undeformed rubber sheet. On each video still, we measured the orientation of the long axis of the Plexiglas ellipse (/3), and the axial ratio of the deformed circle on the rubber sheet (Rf). From Rf, we calculated Q. The deformations we explored with this second suite of experiments were oblique convergence (a = -135°, - 160°, -165°, - 170° and -175°). Experiments varied in speed and scaling. Most experiments took approximately five to ten minutes to run to a finite strain ofRf ~ 2. We performed a few extremely slow experiments that took two to three hours to run to the same finite strain. We discuss scaling in terms of the Argand number (England & McKenzie 1982; Houseman & England 1986), which relates Newtonian viscous flow forces to gravitational body forces. In other words, the Argand number (Ar) reflects the ability of a ductile layer to strain in response to buoyancy forces. In modelling the Earth's deforming lithosphere above an asthenospheric mantle, Ar is approximately 100 for an isostatically compensated crust (S. Lamb, pers. comm. 2001; also see Markley & Tikoff 2003). A reformulation of the Argand number

88

S. GIORGIS ETAL.

Fig. 5. The experimental apparatus. Variation in the speeds of the X and Y motors produces horizontal displacement of the moving plate with respect to the fixed plate. Stretching of the rubber sheet homogeneously distributes deformation into the overlying layers. The apparatus produces convergence, divergence, transcurrence and any degree of oblique motion between the two plates.

(S. Lamb, pers. comm. 2001) for our experiments, which are isostatically uncompensated, is (2)

where p is the density of the fluid (the Earth's asthenosphere or the experiment's silicone), g is acceleration due to gravity, L is the thickness of the fluid layer, 17 is the viscosity of the fluid,

Fig. 6. An example of the first suite of experiments. This is a video still looking down onto a five minute long experiment in oblique divergence (a = 165°). Above the rubber sheet is a 2 cm thickness of clear silicone with two wooden matchsticks embedded into its surface. Rf = 1.4, the matchsticks have rotated to (3 = 105° and 48°. In the final state (Rf = 2.0), the matchsticks are oriented at ft = 120° and 63°. Matchsticks were initially oriented at /3 = 30° and 90°.

CRUSTAL BLOCK ROTATION BY MANTLE FLOW

89

Fig. 7. A video still from the second suite of experiments. The plate in the top part of the image is moving towards the bottom left corner of the image (oblique convergence where a = — 165°). Above the rubber sheet is a 2 cm thickness of clear silicone. In the centre of the rubber sheet is a clear Plexiglas ellipse whose axial ratio is 1.75:1 and whose long axis is oriented at /3 = 132°. The ellipse was originally oriented normal to the fixed plate (/3 — 90°). Visible through the silicone in the upper right corner of the rubber sheet is a strain ellipse, which was inscribed as a circle onto the undeformed rubber sheet. 'Silly putty' walls along the right and left edge of the silicone prevented gravity-driven flow of the silicone during slow experiments.

and e is the strain rate. The fast experiments scale to a low Ar (Ar ^ 1), and the slow experiments scale to a higher Ar (Ar ~ 20). The motors that drive the experimental apparatus are unable to run more slowly and therefore higher Ar numbers could not be obtained. Experimental results and discussion The results of both suites of experiments (Tables 1 & 2) agree well with predictions (Figs 8 & 9). In both suites, we estimate uncertainties of + 1 ° for /3 (the orientation of the rigid object) and ±0.1 for Rf (axial ratio of the finite strain ellipse in the xy plane) from repeated experiments. The most significant source of uncertainty is generated by strain measurements from video stills. Within these uncertainties, matchsticks in oblique convergence/divergence rotate at the same rate as material lines in transpression and transtension (Fig. 8). The Plexiglas plate in oblique convergence rotates at the same rate as Jeffery (1922) and Ghosh & Ramberg (1976) predict for transpression (Fig. 9). Other experiments on Plexiglas ellipses of different axial ratios (1.5:1 and 2:1) and in different transpressional regimes (a = - 175°, -170° and - 160°)

are not reported here, but also show excellent agreement with predictions. Variation of speed in the second suite of experiments strongly suggests that the scaling of these experiments does not affect the results. The slow experiments (for example, the solid circles in Fig. 9b) are gravitationally scaled to the Earth's crust (Lamb 1994). The results of these slow experiments are indistinguishable from the fast experiment (the solid squares and open diamonds in Fig. 9b). Our results are therefore appropriate for comparison to crustal-scale deformation. Furthermore, these results suggest that viscous flow below the Earth's crust is sufficient to drive vertical-axis rotation of isolated rigid crustal blocks. The kinematic relation between such vertical axis rotation and the underlying flow is simple and predictable for zones experiencing convergence, oblique convergence, transcurrence, oblique divergence and divergence.

Application to natural systems In the previous section we demonstrated that the current analytical understanding of the rotation of rigid markers over a viscous medium (Jeffery 1922; Ghosh & Ramberg 1976) is consistent

S. GIORGISCTAL.

90

Table 1. Rotation of single matchsticks Deformation Convergence Convergence Oblique convergence —45° Oblique convergence —45° Oblique convergence — 15° Oblique convergence — 15° Transcurrence Transcurrence Oblique divergence 165° Oblique divergence 165° Oblique divergence 135° Oblique divergence 135° Divergence Divergence Divergence

/3 initial (±1°)

/3 final (±1°)

Q final (±0.05)

/? f (±0.1)

127 89 91 153 33 90 89 42 90 36 92 32 149 91 33

150 90 59 160 21 55 122 66 120 63 111 63 128 91 56

0.48 0.48 0.56 0.56 0.62 0.62 0.65 0.65 0.72 0.72 0.92 0.92 0.94 0.94 0.94

1.9 1.9 2 2 1.9 1.9 1.9 1.9 2 2 2.1 2.1 1.9 1.9 1.9

with data from physical models of a variety of technically realistic settings. While the above modelling is quite simple, it provides an opportunity to test the rates of deformation in a ductile substrate underlying a rotating block (e.g. a bottom-driven system). The rigid upper crustal blocks correspond to the rigid elliptical blocks in the experiments. In a narrow zone surrounding the plate boundary, it is possible that deformation in the basal layer is homogeneous and distributed, similar to the silicone in the physical experiments. Note, however, that the silicone immediately below the rigid clasts may contain complex fabrics (Teyssier et al. 2002). In a bottom-driven model, deformation is transferred from the upper mantle to the rigid upper crust through the ductile lower crust (e.g. Tikoff et al. 2002). The amount of finite rotation that a rigid crustal block has experienced is documented

using palaeomagnetic analysis, as discussed below for specific examples. Shear-wave splitting allows for the determination of the orientation of fabric, interpreted as finite strain, in the upper mantle (e.g. Tommasi et al. 1999). Consequently, we can compare the rotation of the rigid upper crust to that of the lower flow zone (mantle). Shear-wave splitting Seismic anisotropy has been the tool of choice for characterizing mantle deformation in a variety of tectonic settings. This approach is based on the idea that as seismic waves travel through an anisotropic medium they are split into two orthogonal components (shear-wave splitting), analogous to birefringence of light in minerals observed with a petrographic microscope. Once the waves split, they travel at

Table 2. Rotation of Plexiglass ellipse (aspect ratio = 1.75} Deformation Oblique convergence Oblique convergence Oblique convergence Oblique convergence Oblique convergence Oblique convergence Oblique convergence Oblique convergence Oblique convergence Oblique convergence Oblique convergence Oblique convergence

-165° -165° -165° -165° -165° -165° -135° -135° -135° -135° -135° -135°

0 initial (±1°)

j8 final (±1°)

Q final (±0.05)

/? f (±0.1)

90 90 120 120 150 150 90 90 120 120 150 150

128 121 141 147 167 169 113 106 137 139 167 162

0.78 0.64 0.55 0.58 0.6 0.58 0.57 0.49 0.55 0.43 0.48 0.49

2.3 2 1.8 1.9 1.9 1.9 2 1.8 2 1.7 1.8 1.8

CRUSTAL BLOCK ROTATION BY MANTLE FLOW

Fig. 8. Results from the first suite of experiments on matchsticks. Each graph shows the results of a single experiment. Solid lines are the theoretical predictions for the rotation of material lines, and decorated lines are experimental results (matchstick orientation) at fast strain rates (Ar ^ 1). Uncertainties for measurements of matchstick orientation are ± 1°, and uncertainties for the measurement of the axial ratio of the strain ellipse in the horizontal plane (Rf) are +0.1. Coordinate system and theoretical predictions are the same as in Figure 2, but the finite strains are smaller, in accordance with relatively low strains achieved during experiments. In all kinematic regimes here, agreement between matchstick rotations and theoretical predictions is excellent.

different speeds depending on the anisotropy of the material. This difference in velocity creates a time delay between the arrival of the two waves. The magnitude of the total time delay is a function of the degree of anisotropy and thickness of the anisotropic layer. Silver (1996) and Savage (1999) fully discuss the methodology and assumptions involved in the interpretation of seismic anisotropy data. There are several sources of the anisotropy recorded by shear-wave splitting. An overall delay of approximately 1.0-2.Os is commonly observed in neotectonic settings. The crustal component of shear-wave splitting delay times is approximately 0.1-0.2s (Barruol & Kern 1996), and therefore the majority of shear-wave splitting is attributed to the mantle. Volumetrically, olivine is the most abundant mineral in the lithospheric mantle and is mechanically anisotropic with respect to seismic waves (e.g. Christensen 1984). The lattice preferred orientation (LPO) of olivine is presumed to control the shear-wave splitting signal (e.g. Christensen 1984; Nicolas & Christensen 1987). For the purposes of this discussion, we will assume that the orientation of shear-wave splitting tracks the orientation of the long axis of the finite strain ellipsoid for conditions in the uppermost mantle. Ave Lallemant & Carter

91

Fig. 9. Results from some of the second suite of experiments on the Plexiglas ellipse with an axial ratio of 1.75:1. Solid lines are the theoretical predictions for the rotation of rigid ellipses, and symbols are experimental results (orientation of the long axis of the Plexiglas ellipse). Solid squares and open diamonds show data from repeated experiments at fast strain rates (Ar % 1), and solid circles show data from an experiment at slow strain rate (Ar « 25). Uncertainties for measurements of plate orientation are ±1°, and uncertainties for the measurement of the axial ratio of the strain ellipse in the horizontal plane (/?f) are ±0.1. Coordinate system and theoretical predictions are the same as in Figure 4, but the finite strains are smaller. In both regimes of oblique convergence here, agreement between plate rotations and theoretical predictions is excellent. Good agreement between the results of fast and slow experiments shows that scaling to gravity does not affect the outcome of experiments.

(1970) demonstrated that the temperature of deformation affects the activation of different slip systems, which could give rise to other possibilities. However, experimental deformation of olivine, under both high-temperature (Zhang & Karato 1995) and low-temperature (Bystriky et al. 2000) conditions, generally shows that the a-axes of olivine cluster subparallel to the long axis of the finite strain ellipsoid (see also Nicolas & Christensen 1987). The correspondence of the long axis of the finite strain ellipsoid and the direction of shear-wave splitting is also corroborated by numerical models of olivine deformation (e.g. Tommasi et al. 1999) and changing obliquity of shear-wave splitting as one moves away from plate boundaries (e.g. Molnar et al. 1999), in combination with field studies on exposed ultramafic rocks (Nicolas 1989). Distributed mantle deformation in obliquely convergent environments Shear-wave splitting provides a first-order estimate of finite strain in the upper mantle. Numerical

92

S. GIORGIS ETAL

models of transpressional deformation (e.g. Tikoff & Greene 1997) combined with numerical models of olivine deformation (e.g. Tommasi et al 1999) indicate that the long axis of the finite strain ellipsoid lies in the horizontal plane for low angles of oblique convergence. This orientation is critical because large amounts of shear-wave splitting require a horizontal orientation for the clustering of olivine a-axes. The large shear-wave splitting observed (~2 s) in obliquely convergent margins (Trinidad, Russo et al 1996; South Island of New Zealand, Molnar et al 1999) is consistent with this result. Moreover, Little et al (2002) demonstrate that transpression is an appropriate model to explain both crustal deformation and mantle deformation in South Island, New Zealand, although Sutherland et al (2000) argue for a different interpretation of mantle structure. Another key aspect of shear-wave splitting is that it constrains the width of the deforming zone. For a bottom-driven system, a large portion of the mantle must undergo distributed deformation if the overlying crustal blocks undergo vertical-axis rotations. In general, shear-wave splitting is observed over a large area, up to several hundred kilometres perpendicular to the strike of the plate boundary (e.g. Ozalaybey & Savage 1995; Klosko et al 1999; Audoine et al 2000; Polet & Kanamori 2002). Thus, the width of the deformed zone in the mantle is consistent with the mantle controlling large-scale crustal block rotations. Application to vertical-axis rotation to attachment/detachment zones The clutch tectonics model makes predictions on the magnitude and orientation of finite strain in the lithospheric mantle based on the boundary conditions controlling deformation. Therefore, the orientation and magnitude (i.e. length of time delay) of the fast direction of shear-wave splitting should also change based on the boundary conditions of deformation. Assuming a finite rotation of 90° of an upper crustal block in a simple shear regime, one would predict varying orientations and magnitudes of shear-wave splitting in the lower layer (Fig. 1). In a system with a perfect detachment, no finite strain accumulates in the mantle and no shear-wave splitting should be observed. A top-driven (wrench) system deforms only the uppermost lithospheric mantle localized below the block; this leads to very weak or no shear-wave splitting because the magnitude of the time delay depends on the thickness of the anisotropic layer. A perfect

attachment regime yields a stronger anisotropy in the mantle because of the greater amount of finite strain and the thicker layer of deformed material. The strongest observed anisotropy in the mantle layer should result from the bottomdriven end-member case; this is because it requires, relative to the perfect attachment case, a higher amount of finite strain to produce the same finite rotation (y — 77/2 v. y = 77/4; Fig. 1).

Coincidence of upper crustal rotation and mantle deformation A combination of palaeomagnetic data and shear-wave splitting data allows one to determine how the mantle is interacting with the upper crust to rotate crustal blocks. Palaeomagnetic data constrain the finite amount of rotation that a crustal block has experienced. The fast direction of shear-wave splitting constrains the orientation of the finite strain in the upper mantle (Silver 1996). In this section we will apply the results of our model to southern California (Fig. 10) and New Zealand (Fig. 13) in an attempt to gain some insight into the coupled v. non-coupled mantle problem. Eastern Transverse Ranges, southern California The Eastern Transverse Ranges are located approximately 110 miles (175km) east of Los Angeles on the North American plate side of the San Andreas fault (Fig. 10). This area is characterized by a series of strike-slip faults that break up the upper crust into discrete blocks. Using palaeomagnetic data, Carter et al (1987) calculated an average vertical-axis rotation for these blocks of 41° ± 7.7°, which occurred in the last 10 Ma. The local plate motion vector from the Nuvel-la plate model (DeMets et al 1990, 1994) yields a 6° angle of oblique convergence. Although there was a shift in Pacific-North America plate motion in the Late Tertiary, it probably occurred at -8 Ma (Atwater & Stock 1998). Given the uncertainties in the initiation of rotation and the exact plate motion, we assume that the current plate motions were dominant during most of the block rotation. The azimuth of the finite strain ellipse in the upper mantle derived from fast direction of shear-wave splitting data is 103° + 5° (Ozalaybey & Savage 1995; Polet & Kanamori 2002). These data correspond to the LAC station of Ozalaybey & Savage (1995) and are consistent with the CTC, BLA and BC3 stations of Polet & Kanamori

CRUSTAL BLOCK ROTATION BY MANTLE FLOW

93

Fig. 10. Location and tectonic setting of the Eastern Transverse Ranges in southern California. Relative Pacific-North America plate motion vector is from DeMets et al. (1990, 1994). CA, California; NV, Nevada; OR, Oregon. Modified from Carter et al. (1987).

(2002). The stations are all located on the NE side of the San Andreas fault, where the Eastern Transverse Ranges are located. However, all of the stations in southern California are generally subparallel, regardless of their position relative to the San Andreas fault (Polet & Kanamori 2002; P. Davis, pers. comm. 2003). Each fault block was digitized using NIH Image to simplify it to an ellipse, calculate an aspect ratio, and determine the azimuth of the long axis of each ellipse (Table 3). The vertical-axis rotation was removed by rotating the block counterclockwise to obtain an initial orientation. Therefore, we know the aspect ratio of the rigid ellipse, the initial and final azimuths of the long axis of the ellipse, the azimuth of the fault zone, and the kinematics of deformation (6° obliquely convergent transpression). The first analysis assumed a bottom-driven system. Applying these data to the numerical model of rigid clast rotation yields predictions about the orientation of finite strain in the basal driving layer. Using the equations of Ghosh & Ramberg (1976), we forward-model the deformation required to rotate the blocks from their initial to final positions, using the known angle of convergence and a transpressional model as boundary conditions. From the modelled

deformation we calculate the orientation of the long axis of the finite strain ellipse at the rigid ellipse's final position. If the crustal blocks are rotating as part of a bottom-driven system, the predicted long axis of finite strain should correspond to finite strain in the mantle (Fig. 11). In southern California, the azimuth of the long axis of the finite strain ellipse is predicted to be in the range of 102° to 110° (Table 3). This range of predictions reflects the propagation of palaeomagnetic uncertainties through the calculations. The orientation of the observed mantle fabric (using shear-wave splitting) and the predicted mantle fabric from rigid-body rotations agree surprisingly well, if a bottom-driven system is assumed (Fig. 11). In fact, the observed splitting data fall exactly between predictions made from palaeomagnetically derived crustal block rotations. The second analysis assumed a side-driven system. The difficulty with this approach is related to the geometry of an elliptical block in a dominantly transcurrent zone: the shear zone width varies depending on whether the long or short axis of the ellipse is oriented parallel to the boundary. Therefore, to compare the sidedriven model quantitatively relative to the bottom-driven model, our calculations assume a

94

S. GIORGIS ETAL.

Table 3. Data and results for individual crustal blocks Location

Southern California* North Central South New Zealand1"

Aspect ratio

4 3 3.6 1 1.5

Clockwise palaeomagnetic rotation

4.1° 4.1° 4.1° 29° 29°

+ + + + ±

7.7° 7.7° 7.7° 6.3° 6.3°

Azimuth of fast direction of shear-wave splitting

103° 103° 103° 046° 046°

+ + ± + ±

5° 5° 5° 7° 7°

Azimuth of predicted long axis of finite strain

106° 105° 105° 036° 034°

+ + + + ±

4° 3° 3° 3° 3°

* Palaeomagnetic data from Carter et al. (1987); shear-wave splitting data from Ozalaybey & Savage (1995) consistent with Polet & Kanamori (2002). Palaeomagnetic data from Roberts (1992) and Little & Roberts (1997); shear-wave splitting data from Audoine et al. (2000) consistent with Klosko et al (1999).

circular crustal block (Fig. 12). The rates of rotation vary for different angles of convergence and elliptical ratios. Regardless, a clear pattern emerges, that can be thought of in two equivalent ways: (1) the side-driven system always rotates more for a given amount of finite strain in the basal layer; or (2) the side-driven system rotates more quickly with respect to applied transcurrent motion. For any amount of palaeomagnetic rotation, the predicted orientation of finite strain in the ductile layer is always closer to the shear zone boundary for the

Fig. 11. Comparison of model predictions of the azimuth of the long axis of finite strain in the mantle (outlined in heavy black lines) and the azimuth of the fast direction of shear-wave splitting (grey shaded region). Dashed line is the orientation of the San Andreas fault (a) or the Alpine fault (b). Data from the central block are representative of all the crustal blocks examined in southern California (Table 3).

bottom-driven case (Fig. 12). For the boundary conditions of the San Andreas fault system example, the predictions of the side-driven v. bottom-driven models are separated by ~8°. The range in predictions is based on the error associated with the palaeomagnetic data. As shear-wave splitting datasets have an error of + 5°, distinguishing between the bottom-driven v. side-driven systems is possible. Note also that this second model for shear-wave splitting data does not accurately describe the kinematics of the San Andreas fault system, because simple shear deformation was assumed. However, the difference in rotation rates between simple shear deformation and the actual angle of oblique convergence (a = 6°) is minimal.

Marlborough fault system, South Island, New Zealand Applying the same methods, we can examine the rotation of a crustal block in the Marlborough fault system (Fig. 13), which has the advantage of a larger shear-wave splitting dataset. However, the dimensions of the crustal block are less well defined because much of the block is offshore. Little & Roberts (1997) delineate a line separating zones that have experienced vertical-axis rotation from those that have not (Fig. 13). Owing to the limited constraints on the dimensions of this block, a range of aspect ratios is presented (Table 3). If the rotating block includes the area up to the intersection of the ancestral Alpine fault and the western Pacific subduction zone, an aspect ratio of 1.5 is obtained. Otherwise, the block is nearly equant (i.e. aspect ratio = 1.0). Palaeomagnetic data (Roberts 1992; Little & Roberts 1997) indicate that the block has rotated an average of 29° + 6.3° clockwise in the last 4 Ma. The local plate motion vector

CRUSTAL BLOCK ROTATION BY MANTLE FLOW

95

Fig. 12. (a,b) A hypothetical scenario similar to the southern California example. An equant crustal block is rotated 41.1° + 7.7° in either a side-driven (a) or a bottom-driven (b) transcurrent system. The orientation of the fast direction of shear-wave splitting should be parallel to the long axis of the finite strain ellipsoid (the dark ellipse). Side-driven v. bottom-driven boundary conditions predict different orientation (c) and magnitudes of shear-wave splitting (a,b). (d) Predictions for 29° + 6.3° of rotation in a simple shear system, similar to the New Zealand example. Dashed line is the orientation of the San Andreas fault (c) or the Alpine fault (d).

(DeMets et al 1990, 1994) gives a 20° angle of oblique convergence on the Alpine fault. Similar to the California example, the rotating block was simplified to an ellipse (Table 3). The first model assumed a bottom-driven system. A forward model was calculated to rotate the block from its initial to final position. A range of 033° to 039° is calculated for the azimuth long axis of the finite strain ellipse in the upper mantle (Table 3). The azimuth of the fast direction shear-wave splitting data (Klosko et al. 1999; Audoine et al 2000) is 046° ± 7°. These shear-wave splitting data are consistent with, but slightly clockwise of, the azimuths predicted by the model (Fig. 11). The second analysis assumed a side-driven system. Because the block was approximately circular, our calculations that assume a circular crustal block (Fig. 12) are appropriate. However, as noted above, this analysis also requires the assumption of simple shear, which does not accurately describe the kinematics of the Alpine fault system (a = 20°). Regardless, the

predictions of the side-driven model suggest a larger (~8°) obliquity between the plate margin and the finite strain in the ductile substrate. These data are not consistent with the observed shear-wave splitting, even within uncertainties bars. Therefore, given the assumptions of the analysis, side-driven boundary conditions are not possible for this region. Discussion Side-driven v. bottom-driven systems The coincidence of upper crustal and mantle deformation has been recognized in a variety of neotectonic settings (e.g. Holt 2000; Little et al 2002). Holt (2000) noted the concurrence of geodetic strain data and shear-wave splitting data in Tibet. Little et al (2002) used two offset markers to calculate finite strain in the upper crust in the South Island of New Zealand, and noted its coincidence with upper mantle deformation recorded by shear-wave splitting. Also using

96

S. GIORGIS ETAL.

Fig. 13. Location and tectonic setting of the Marlborough fault system on South Island, New Zealand. The on-shore boundaries of the rotating block are from Little & Roberts (1997). The off-shore boundaries of the rotating block are uncertain and therefore a range of possible aspect ratios for this crustal block are presented. Relative Pacific-Australian plate motion vector is from DeMets et al (1990, 1994). Modified from Little & Roberts (1997).

the South Island of New Zealand, Moore et al. (2003) were able to predict the orientation of shear-wave splitting by integrating 6.5 Ma of motion occurring at the present geodetic rate. An assumption of this analysis was that the surface deformation was directly related to deformation lower in the lithosphere. Regardless of the exact relation, the presence of measurable shear-wave splitting allows us to eliminate the perfect detachment and top-driven models because they are inconsistent with the development of a strong mantle fabric. This leaves two options for the interpretation of crustal deformation above a highly anisotropic mantle: either (1) the upper crust and the lithospheric mantle had the same boundary conditions and thus deformed in similar ways (side-driven system; e.g. Holt 2000); or (2) deformation of the upper crust is controlled by flow of the lithospheric mantle (bottom-driven system; e.g. Molnar 1992; Teyssier & Tikoff 1998). Both examples provided in this paper - South Island of New Zealand and southern California - are better matched by the bottom-driven model. Predictions based on the crustal blocks in southern California, assuming a bottomdriven system, are exactly consistent with the

observed fast direction of shear-wave splitting (Table 3; Fig. 11). The mantle fabric predicted by a side-driven system would be at a higher angle to the plate boundary than observed. The case for bottom-driven rotation is even stronger in New Zealand. In this setting, the observed mantle fabric is oriented slightly closer to the plate boundary (defined by the Alpine fault) than the fabric orientation predicted by upper crustal rotations if a bottom-driven system is assumed (Table 3; Fig. 11). The predicted orientation of mantle fabric in a side-driven system is at a higher angle to the plate boundary than the prediction for a bottom-driven system (Fig. 12d), and therefore not consistent with the observations. Consequently, within the limitations of the assumptions (see below), the analysis suggests that these systems are bottom-driven. This is further supported by geodetic surveys. In both regions, geodetic surveys indicate that the upper crustal deformation zones are broadly continuous (e.g. Beavan & Raines 2001, for the South Island of New Zealand). This occurs despite an extremely complicated array of rotational (transrotational) blocks with a variety of fault orientations, particularly in southern California (Dickinson 1996). If

CRUSTAL BLOCK ROTATION BY MANTLE FLOW

side-driven motion were the general case, one would expect transient motion on faults as they become increasingly better aligned for slip. The consistency of motion, despite the heterogeneity of structure, suggests that the upper crustal motion is dependent on motion lower in the lithosphere where broad zones of deformation are more likely. Assumptions and applicability of the model Applying our model to actively deforming plate boundaries requires three assumptions: (1) no block-block interaction; (2) an isotropic mantle (i.e. had no fabric) prior to the deformation that drove the block rotation; and (3) deformation was steady-state from a kinematic perspective. Block-block interactions would have the tendency to slow down the rotation rate of crustal blocks, for a given strain in the ductile substrate. Therefore, if the effect was very large, a side-driven system may begin to appear as a bottom-driven system. Although there are a few studies of block interactions in these systems (e.g. Lamb 1987, 1994; Markley & Tikoff 2003), the strength of this effect is unknown. Another problem with any interpretation is the possibility of inherited mantle fabric (i.e. mantle fabric that developed prior to upper crustal rotational deformation). The age of rotation of the block in New Zealand is less than 4 Ma (Little & Roberts 1997), while the Alpine fault has been active for at least 25 Ma (Cooper et al 1987). The Pacific-Australian plate boundary has a much longer history (Sutherland 1999). This history indicates that the mantle was potentially not isotropic and developed some fabric prior to the deformation associated with the rotation of the upper crustal block. If this component could be quantified, it may lead to even better agreement between the bottom-driven model and the shear-wave splitting data. The last problem involves the assumption of steady-state kinematics. Because of a change of motion of the Pacific plate in the Late Miocene, both areas may have experienced a change in boundary conditions during verticalaxis rotation. However, the effect is more relevant to the southern California example, since the rotation has occurred since 10 Ma. Although the rate of plate motion remained approximately equivalent, there was a change in motion from slightly obliquely divergent (prior to ~8 Ma; Atwater & Stock 1998) to obliquely convergent. Consequently, the transcurrent component of motion, which is dominantly responsible for the rotation, would have been very similar in both

97

cases. If better data on the timing of rotation were available and more exact plate motion models were made, it would be possible to incorporate non-steady-state motion into the deformation models. Another important limitation to the model is the assumption of rigidity. While most obliquely convergent or obliquely divergent plate boundaries demonstrate some rotational component of deformation, the assumption of rigidity is not universal. For example, deformation in the San Andreas fault system north of the Garlock fault is characterized by strike-slip partitioning (e.g. Teyssier & Tikoff 1998). In this case, the wrench borderlands of the San Andreas fault are both translating and internally deforming. The fundamental change in behaviour between the different styles of deformation apparently occurs at the Garlock fault, which has acted as a major boundary between different tectonic behaviour since the Late Mesozoic. It is also interesting to note that a change in shear-wave splitting behaviour also occurs at this boundary (Ozalaybey & Savage 1995; Polet & Kanamori 2002). To the north, the location of the San Andreas fault affects the fabric in the mantle, whereas it does not to the south. The results of our modelling for California and New Zealand are consistent with a bottom-driven system. Consequently, it appears that the deformation in the mantle is the principal cause of vertical-axis rotations of large crustal blocks in these settings. Admittedly, we are utilizing a sparse dataset and we cannot conclusively rule out the effects of a pre-existing anisotropy prior to block rotation. Future application of this methodology will help constrain the nature of crustmantle interaction and can help distinguish between side-driven and bottom-driven systems. Conclusions The vertical-axis rotation of rigid crustal blocks potentially allows one to distinguish whether upper crustal deformation is bottom-driven or side-driven. Using models of transpressional and transtensional kinematics for particular elliptical ratios of vertically rotating blocks, it is possible to calculate the associated strain of a ductile substrate. For a given amount of rotation, bottomdriven deformation predicts higher strain oriented at a lower angle to the boundary than side-driven deformation. We apply this approach to southern California and New Zealand by using documented palaeomagnetic rotations and plate motion vectors, and calculate the strain recorded by the material below rigid blocks. These strain calculations are

98

S. GIORGIS ETAL.

compared to shear-wave splitting data, which record upper mantle fabric, from the same region. These data suggest that similar deformation is recorded by the upper crust and lithospheric mantle. A bottom-driven flow, in which mantle deformation drives upper crustal rotations, is most consistent with these observations. Thanks are due to Simon Lamb and Bruce Luyendyk for comments on an earlier version of some of this manuscript. We thank Tim Little for discussions on the Marlborough fault system and New Zealand in general. Paul Davis is thanked for a helpful discussion of shear-wave splitting in southern California. The formal reviews of Phillip England and an anonymous reviewer improved the manuscript. We thank Pierre Bouihol for his assistance with the physical experiments and Eric Horsman for a thoughtful review. Maitri Venkat-Ramani provided the engineer's drawing of the experimental apparatus. SG and BT acknowledge support from NSF grant EAR 0001092 and a Packard Foundation Fellowship to BT. MM acknowledges support from Mount Holyoke College and the American Association of University Women.

References ARBARET, L., FERNANDEZ, A., JEZEK, J., ILDEFONSE, B., LAUNEAU, P. & DIOT, H. 2000. Analogue and numerical modeling of shape fabrics: applications to strain and flow determination in magmas. Transactions of the Royal Society of Edinburgh, 90, 97-109. ATWATER, T. & STOCK, J. 1998. Pacific-North America plate tectonics of the Neogene southwestern United States: an update. In: ERNST, W.G. & NELSON, C.A. (eds) Integrated Earth and Environmental Evolution of the Southwestern United States. Geological Society of America, 393-420. AUDOINE, E., SAVAGE, M.K. & GLEDHILL, K. 2000. Seismic anisotropy from local earth quakes in the transition region from a subduction to a strikeslip plate boundary, New Zealand. Journal of Geophysical Research, 105, 8013-8033. AVE LALLEMANT, H.G. & CARTER, N.L. 1970. High temperature flow of dunite and peridotite. Geological Society of America Bulletin, 81, 2181-2202. BARRUOL, G. & KERN, H. 1996. Seismic anisotropy and shear-wave splitting in lower-crustal and upper-mantle rocks from the Ivrea Zone; experimental and calculated data. In: MAINPRICE, D. & VAUCHEZ, A. (eds) Dynamics of the subcontinental mantle, from seismic anisotropy to mountain building. Physics of the Earth and Planetary Interiors, 95, 175-194. BEAVAN, J. & RAINES, J. 2001. Contemporary horizontal velocity and strain rate fields of the Pacific-Australian plate boundary zones through New Zealand. Journal of Geophysical Research, 106,741-770. BYSTRIKY, M., KUNZE, K., BURLINI, L. & BURG, J. 2000. High shear strain of olivine aggregates; rheological and seismic consequences. Science, 290, 1564-1567.

CARTER, J.N., LUYENDYK, B.P. & TERRES, R.R. 1987. Neogene clockwise tectonic rotation of the eastern Transverse Ranges, California, suggested by paleomagnetic vectors. Geological Society of America Bulletin, 98, 199-206. CHRISTENSEN, N.I. 1984. The magnitude, symmetry and origin of upper mantle anisotropy based on fabric analyses of ultramafic tectonites. Geophysical Journal of the Royal Astronomical Society, 76, 89-111. COOPER, A.F., BARREIRO, B.A., KIMBROUGH, D.L. & MATTINSON, J.M. 1987. Lamprophyre dike intrusion and the age of the Alpine Fault. Geology, 15, 941-944. DEMETS, C, GORDON, R.G., ARGUS, D.F. & STEIN, S. 1990. Current plate motions. Geophysical Journal International, 101, 425-478. DEMETS, C., GORDON, R.G., ARGUS, D.F. & STEIN, S. 1994. Effect of recent revisions to the geomagnetic time scale on estimates of current plate motions. Geophysical Research Letters, 21, 2191-2194. DEWEY, J.F., HOLDSWORTH, R.E. & STRACHAN, R.E. 1998. Transpression and transtension zones. In: HOLDSWORTH, R.E., STRACHAN, R.A. & DEWEY, J.F. (eds) Continental Transpressional and Transtensional Tectonics. Geological Society, London, Special Publications, 135, 1-14. DICKINSON, W.R. 1996. Kinematics of Transnational Tectonism in the California Transverse Ranges and its Contribution to Cumulative Slip Along the San Andreas Transform Fault System. Geological Society of America Special Paper 305. ENGLAND, P.C. & MCKENZIE, D.P. 1982. A thin viscous sheet model for continental deformation. Geophysical Journal of the Royal Astronomical Society, 70, 295-321. FOSSEN, H., TIKOFF, B. & Teyssier, C. 1994. Strain modeling of transpressional and transtensional deformation. Norsk Geologisk Tidskrift, 74, 134-145. FREEMAN, B. 1985. The motion of rigid ellipsoidal particles in slow flows. Tectonophysics, 113, 163-183. GHOSH, S.K. & RAMBERG, H. 1976. Reorientation of inclusions by combination of pure and simple shear. Tectonophysics, 34, 1 —70. GIORGIS, S. & TIKOFF, B. 2004. Constraints on kinematics and strain from feldspar porphyroclast populations. In: ALSOP, I., HOLDSWORTH, R., MCCAFFERY, K. & HAND, M. (eds) Faults and Flow Processes in Shear Zones. Geological Society of London, Special Publications, 224, 265-285. HOLT, W. 2000. Correlated crust and mantle strain fields in Tibet. Geology, 28, 67-70. HOUSEMAN, G. & ENGLAND, P. 1986. Finite strain calculations of continental deformation: method and general results for convergent zones. Journal of Geophysical Research B3, 91, 3651-3663. JEFFERY, G.B. 1922. The motion of ellipsoidal particles immersed in a viscous fluid. Proceedings of the Royal Society of London Series A, 102, 161 — 179. JEZEK, J., MELKA, R., SCHULMANN, K. & VENERA, Z. 1994. The behavior of rigid triaxial ellipsoidal particles in viscous flows - modeling of fabric

CRUSTAL BLOCK ROTATION BY MANTLE FLOW evolution in a multiparticle system. Tectonophysics, 299, 165-180. JEZEK, J., SCHULMANN, K. & SEGETH, K. 1996. Fabric evolution of rigid inclusions during mixed coaxial and simple shear flows. Tectonophysics, 251, 203-221. KLOSKO, E.R., Wu, F.T. et al 1999. Upper mantle anisotropy in the New Zealand region. Geophysical Research Letters, 26, 1497-1500. LAMB, S.H. 1987. A model for tectonic rotations about a vertical axis. Earth and Planetary Science Letters, 84, 75-86. LAMB, S.H. 1994. Behavior of the brittle crust in wide plate boundary zones. Journal of Geophysical Research B3, 99, 4457-4483. LITTLE, T.A. & ROBERTS, A.P. 1997. Distribution and mechanism of Neogene to present-day vertical axis rotations, Pacific-Australian plate boundary, South Island, New Zealand. Journal of Geophysical Research, 102, 20447-20468. LITTLE, T.A., SAVAGE, M.K., & TIKOFF, B. 2002. Relationship between crustal finite strain and seismic anisotropy in the mantle, Pacific-Australia plate boundary zone, South Island, New Zealand. Geophysical Journal International, 151, 160-169. MARKLEY, M. & TIKOFF, B. 2003. Matchsticks on parade: vertical axis rotation in oblique convergence/divergence. Journal of Geophysical Research, 107, doi: 10.1029/2002JB001826. MCKENZIE, D. & JACKSON, J. 1983. The relationship between strain rates, crustal thickening, paleomagnetism, finite strain, and fault movements within a deforming zone. Earth and Planetary Science Letters, 65, 182-202. MOLNAR, P. 1992. Brace-Goetze strength-profiles, the partitioning of strike-slip and thrust faulting at zones of oblique convergence, and the stressheat flow paradox of the San Andreas fault. In: EVANS, B. & WONG, T.-F. (eds) Fault Mechanics and Transport Properties of Rocks. Academic Press, New York, 435-459. MOLNAR, P., ANDERSON, H., et al. 1999. Continuous deformation versus faulting through the continental lithosphere of New Zealand. Science, 286, 516-519. MOORE, M., ENGLAND, P. & PARSONS, B. 2003. Relation between surface velocity field and shearwave splitting in the South Island of New Zealand. Journal of Geophysical Research, 107, doi: 10.1029/2000JB000093. NICOLAS, A. 1989. Structures of Ophiolites and Dynamics of Oceanic Lithosphere. Kluwer Academic, Dordrecht. NICOLAS, A. & CHRISTENSEN, N.I. 1987. Formation of anisotropy in upper mantle peridotites. Geodynamics Series, 16, 111 -123. OZALAYBEY, S. & SAVAGE, M.K. 1995. Shear-wave splitting beneath the western United States in relation to plate tectonics. Journal of Geophysical Research, 100, 18135-18149. PASSCHIER, C.W. 1997. The fabric attractor. Journal of Structural Geology, 19, 113-127. POLET, J. & KANAMORI, H. 2002. Anisotropy beneath California: shear wave splitting measurements

99

using a dense broadband array. Geophysical Journal International, 149, 313—327. RAMBERG, H. 1975. Particle paths, displacement, and progressive strain applicable to rocks. Tectonophysics, 28, 1-37. ROBERTS, A.P. 1992. Paleomagnetic constraints on the tectonic rotation of the southern Hikurangi margin, New Zealand. New Zealand Journal of Geology and Geophysics, 35, 311-323. Russo, R.M., SILVER, P.O., FRANKE, M., AMBEH, W.G. & JAMES, D.E. 1996. Shear-wave splitting in northeast Venezuela, Trinidad, and the eastern Caribbean. Physics of the Earth and Planetary Science Interiors, 95, 251-215. SANDERSON, D.J. & MARCHINI, W.R.D. 1984. Transpression. Journal of Structural Geology, 6,449-458. SAVAGE, M.K. 1999. Seismic anisotropy and mantle deformation: what have we learned from shear wave splitting? Reviews of Geophysics, 37, 65-106. SILVER, P.O. 1996. Seismic anisotropy beneath the continents: probing the depths of geology. Annual Reviews of Earth and Planetary Sciences, 24, 385-432. SUTHERLAND, R. 1999. Cenozoic bending of New Zealand basement terranes and Alpine Fault displacement: a brief review. New Zealand Journal of Geology and Geophysics, 42, 295-301. SUTHERLAND, R., DAVEY, F. & BEAVAN, J. 2000. Plate boundary deformation in South Island, New Zealand, is related to inherited lithospheric structure. Earth and Planetary Science Letters, 111, 141-151. TEYSSIER, C. & TIKOFF, B. 1998. Strike-slip partitioned transpression of the San Andreas fault system: a lithospheric-scale approach. In: HOLDSWORTH, R.E., STRACHAN, R.A. & DEWEY, J.F. (eds) Continental Transpressional and Transtensional Tectonics. Geological Society, London, Special Publications, 135, 143-158. TEYSSIER, C., TIKOFF, B. & WEBER, S. 2002. Attachment between brittle and ductile crust at wrenching plate boundaries. EGS Stephan Mueller Special Publication Series, I, 119-144. TIKOFF, B. & GREENE, D. 1997. Stretching lineations in transpressional shear zones: Journal of Structural Geology, 19, 29-40. TIKOFF, B. & TEYSSIER, C. 1994. Strain modeling of displacement-field partitioning in transpressional origins. Journal of Structural Geology, 16, 15751588. TIKOFF, B. TEYSSIER, C. & WATERS, C. 2002. Clutch tectonics and the partial attachment of lithospheric layers. EGS Stephan Mueller Special Publication Series, 1,93-117. TOMMASI, A., TIKOFF, B. & VAUCHEZ, A. 1999. Upper mantle tectonics: three-dimensional deformation, olivine crystallographic fabrics and seismic properties. Earth and Planetary Science Letters, 168, 173-186. VENKAT-RAMANI, M. & Tikoff, B. 2002. Physical models of transtensional folding. Geology, 30, 523-526.

100

S. GIORGIS ETAL.

WILLIS, D.G. 1977. A kinematic model of preferred orientation. Bulletin of the Geological Society of America, 88, 883-894.

ZHANG, S. & KARATO, S.I. 1995. Lattice preferred orientation of olivine aggregates deformed in simple shear, Nature, 375, 774-777.

Strain gradients in transpressional to transtensional attachment zones C. TEYSSIER & L. CRUZ Department of Geology and Geophysics, University of Minnesota, Minneapolis, MN 55455 USA Abstract: Attachment zones couple the rheological layers of lithosphere. In wrench settings, attachment zones accommodate the transition from relatively continuous wrenching at depth to discrete strike-slip faulting of rigid blocks in the upper crust. Strain is controlled by a component of wrench shearing as well as a component of horizontal shearing associated with the differential displacement of finite-width rigid blocks. Strain modelling of wrench attachments predicts high lateral and vertical strain gradients and specific foliation patterns showing antiforms and funnel-shaped synforms. Lineations are shallowly plunging and oriented close to the direction of wrenching. Shear sense reverses across the vertical axial surfaces of synforms and antiforms. In transpression and transtension attachments developed during low-angle oblique convergence or divergence, the pattern of foliation and lineation is similar to that produced in wrench attachments. Transpression attachments display gradients in the shape of the finite strain ellipsoid, from flattening at the base to strongly constrictional beneath the rigid blocks, owing to the increased effect of the horizontal shear component. Conversely, transtension attachments show constriction at the base changing to flattening beneath the rigid blocks. The location of this fabric change within attachment zones is insensitive to finite displacement and angle of convergence or divergence, and therefore should be one of the most robust criteria to identify transpression and transtension attachments. In general, the component of coaxial flow that characterizes transpressional and transtensional systems decreases upward through attachment zones, due to the increased role of the horizontal simple shear in the finite vorticity. These strain and kinematic gradients are a robust result of attachment modelling and can be used as indicators of attachments developed in wrench, transpression, or transtension.

According to experimental work and field studies, the continental lithosphere consists of subhorizontal rheological layers that include the brittle upper crust, the ductile crust and the lithospheric mantle (Kohlstedt et al 1995; Vauchez et al. 1998). In some tectonic settings, lithospheric layers have a tendency to delaminate or slide relative to one another. In regions of lithospheric contraction, fold-thrust wedges form above decollements, and in regions of lithospheric extension, low-angle detachments develop. In wrench and obliquely convergent/ divergent tectonic settings, where the zone of deformation is roughly orthogonal to the lithospheric layering, the lithospheric layers appear to be mechanically coupled (England & Wells 1991; Molnar 1992), implying the existence of 'attachment,' rather than 'detachment' zones (Teyssier et al. 2002; Tikoff et al. 2002). The motion of rigid blocks (i.e. upper crust or lithosphere) displaced along strike-slip faults imparts velocity gradients, and therefore strain,

between the rigid blocks and the underlying layer that may be shearing more continuously, The aim of this chapter is to investigate the types of fabrics and structures that may develop as a result of attachment tectonics, starting with the simple case of wrench zones and moving to obliquely convergent and divergent systems, Attachment tectonics Typically, where continental crust is involved in wrenching, the zone of upper crustal deformation is diffuse (Molnar 1988), commonly spread over widths of 100km or more. In the upper crust, strike-slip faults separate crustal blocks that translate, rotate and may contract or extend to accommodate the component of oblique convergence or divergence. These upper crustal blocks move relatively rigidly as shown by geological studies and geodetic surveys (Bourne et al. 1998&; Savage ef al. 1999): velocity gradients are concentrated near the strike-slip faults. Sylvester

From: GROCOTT, J., MCCAFFREY, K. J. W., TAYLOR, G. & TIKOFF, B. (eds) 2004. Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 101-115. 0305-8719/04/$15 © The Geological Society of London 2004.

102

C. TEYSSIER & L. CRUZ

(1988) addressed the deformation below these rigid blocks in terms of the 'mechanical stratigraphy' of wrench zones. In principle, two end-members can be considered: (1) the strike-slip faults extend to great depth and accommodate all of the wrench motion (Vauchez & Tommasi 2003); or (2) at some depth, wrench motion is distributed in a broad zone of deformation (Bourne et al 1998a). The transition between the layer of discrete faulting and the layer of continuous flow may take various forms. Following Bourne et al (19980, b), we consider the case where the two layers are 'attached' by a transition zone, or attachment zone (Fig. 1) (Teyssier et al 2002; Tikoff et al 2002). The attachment zone maintains strain continuity between the rigid blocks above and the simple shear zone below. A range of possibilities obviously exists between detachment and attachment, depending on how much coupling exists across the zone (partial attachment of Tikoff et al 2002). The concept of attachment is useful because it forces

one to consider the velocity gradients, the instantaneous strain and the finite strain that could develop beneath the rigid blocks. Bourne et al (1998a, b) discussed the following conceptual framework (Fig. 1). Rigid blocks of upper crust are translated by strike-slip faulting; at depth, deformation is accommodated by continuous simple shear. The thickness of the attachment zone depends on the width of the overlying rigid blocks and the rheology of the ductile crust in which the attachment develops. If the ductile crust behaves as a Newtonian fluid, a solution to Laplace's equations shows that the velocity field in a stack of horizontal planes in the attachment zone is sinusoidal (Fig. Ib), with the amplitude of the sinusoid decreasing as the velocity field approaches that of simple shear at depth (Bourne et al 1998£). For a block of width a (for example 20 km), the deformation zone in a Newtonian fluid would be ~a/ir (~6km) thick (Bourne et al 19980).

Fig. 1. (a) Conceptual framework for an attachment zone that accommodates the strain between rigid blocks or crustal slivers above, and a homogeneous simple shear zone below, (b) Velocity field for rigid blocks (thick line), homogeneous simple shear at depth (straight solid line), and attachment zone (sinusoidal dashed lines), (c) In a fourlayer attachment zone - levels A (bottom) to D (top) - displacement is approximated by a linear variation of the horizontal shear induced by the relative displacement of rigid blocks; the horizontal shear increases upward and outward from the centre of the blocks. Modified from Bourne et al (1998«).

STRAIN GRADIENTS IN ATTACHMENT ZONES

For a non-Newtonian fluid of power-law rheology, the predicted thickness of the deforming zone is less than a/ TT because deformation tends to localize, and the thickness of the attachment zone may vary laterally as a function of strain rate (Teyssier et al 2002). In this chapter we use thicknesses of a/4 and «/8 for attachment zones of uniform thickness. The attachment zone is ~5 km and ~2.5 km thick beneath blocks that are ~20 km wide. In principle, the same principles would hold for rigid lithospheric blocks, like in Turkey or SE Asia, that are of the order of a few hundred kilometres wide; in this case, the attachment zone beneath the rigid blocks of lithosphere would be of the order of 50 km thick. In a previous paper, Teyssier et al. (2002) considered the problem of rotation and translation of upper crustal rigid blocks in a wrench zone; in this chapter, we explore more specifically the finite strain gradients that develop within attachment zones during the translation of rigid blocks/slivers in wrench systems. In addition, this analysis is extended to attachment zones developed in transpression and transtension. Modelling of strain gradients On the basis of strain continuity in attachment zones, kinematic models were constructed using appropriate sets of boundary conditions (Fig. 2). The boundary conditions for attachments developed in wrenching include a wrench simple shear and a horizontal simple shear (Figs 2 & 3). Attachments developed in transpression and transtension deform according to those same boundary conditions and also in response to a pure shear component that shortens the attachment zone in a horizontal and vertical direction, respectively (Fig. 2). In this study, strain is calculated for discrete elements in attachment zones. Deformation is considered constant-volume and homogeneous in each element. The modelling uses the deformationmatrix approach published in Tikoff & Fossen (1993) and Fossen & Tikoff (1993) and a version of the 3D-STRAIN program (Tikoff & Fossen 1996). 3D-STRAIN uses an upper triangular deformation matrix, assuming no external 'spin' component of deformation, e.g. no external rotation of coordinate axes. The deformation matrix in its general form is shown in Figure 2, and is characterized by £1? k2 and &3, the coaxial components applied along jc, y and z, respectively. F^, Yxz and Yyz are the simple shear components representing a wrench shear in the x direction, hereafter called y(w), a horizontal shear in the x direction, hereafter called y(hs), and a horizontal shear in the y direction

103

Fig. 2. Components of deformation used in the deformation matrix. Wrench attachment is described by the combination of wrench y(w) and horizontal simple shear y(hs), with shear direction parallel to the ;c-axis; transpression and transtension require a component of pure shear, with k\ = 1, k2 < 1, £3 = l/k2 for transpression and k} = 1, k2 > 1, k3 = l/k2 for transtension, under constant-volume conditions.

(here Tyz = 0). The wrench, transpression and transtension attachment models developed in this paper use the deformation matrix to superpose the two orthogonal simple shears y(w) and y(hs); in addition, transpression is achieved when [&2 < 1 and k3 = 1/&2L and transtension is modelled with [k2 > 1 and k3 = l/&2], with k[ = 1 in all of the cases studied herein (Fig. 2). Given these boundary conditions, exact strain solutions are derived for the shape and orientation of the finite strain ellipsoid and finite kinematic vorticity for each discrete element in the modelled attachment zones. These results serve as a basis for predicting the orientation of planar and linear fabrics, and the variations in finite strain and vorticity across potential attachment zones developed in nature. This chapter presents two types of models: (1) attachments developed in pure wrenching (see also Merle & Gapais 1997; Teyssier et al. 2002); and (2) attachments developed in transpression and transtension. For each case, special attention is given to the strain gradients that develop in these types of attachment zones. Although this chapter does not present an exhaustive

104

C. TEYSSIER & L. CRUZ

Fig. 3. Block diagram showing attachment zone in wrench setting; strain in attachment zone accommodates wrench shear y(w) as well as a component of horizontal shear y(hs) induced by the displacement of rigid blocks. In order to calculate strain, the attachment zone is discretized as shown, and the horizontal shear component is varied linearly both laterally and vertically through the attachment zone. The symmetry about the centre of the attachment zone dramatically reduces the number of solutions to be calculated (shown in grey area). For the case shown, the wrench shear strain below the attachment zone is y = 1. Note the location of the series of strip maps shown in Figure 4.

series of results, the most robust aspects of strain solutions, which are somewhat independent of amount of displacement and relatively insensitive to angle of convergence and divergence, provide first-order criteria to help recognize attachment zones as such, if indeed they exist in nature.

Wrench attachments To map the lateral and vertical strain gradients in detail, the wrench attachment zone is discretized and treated as domains of homogeneous deformation. The attachment zone is divided into four layers, from layer A at the base to layer D

STRAIN GRADIENTS IN ATTACHMENT ZONES

at the top. Beneath given rigid blocks, each horizontal level of the attachment zone is then divided equally into nine sections, including a central domain and four domains on each side that are the mirror images (Fig. 3). This symmetry significantly reduces the number of solutions to be calculated. During pure wrenching, an attachment zone is sandwiched between a wrench simple shear zone below and a rigid block above (Fig. 3). The wrench shear imparts a wrench component of deformation to the attachment zone. In most of the models presented here, this component is assumed to be constant throughout the attachment zone; this is equivalent to the attachment zone responding to a side boundary condition of a homogeneous wrench simple shear. However, in principle, a wrench shear gradient could develop beneath the strike-slip fault between two rigid blocks. The attachment zone is thinner than it is wide by a factor of at least 77, as discussed above, and therefore the wrench gradient related to the strike-slip fault is likely localized beneath the fault zone. In the following modelling, this sort of wrench shear gradient is explored for one specific example of a wrench attachment zone. The presence of the rigid blocks imparts a variable component of horizontal shear controlled by the difference in velocity between the wrench zone and the rigid blocks (Fig. Ib). In the intervening attachment zone, this horizontal shear component is described by a horizontal shear plane and a shear direction parallel to the strike-slip faults. Consider level D, the upper level of the attachment zone (Fig. 3). In the centre of the rigid upper block, the velocity of particles is the same as in the deep shear zone beneath; therefore, there is no horizontal shear component in this middle column of the attachment zone where the strain is controlled solely by the wrench component. Away from the centre of the block, the component of horizontal shear increases toward a maximum beneath the edges of the upper blocks (Fig. 3). Similarly, the component of horizontal shear varies vertically; it is zero in the wrench zone and increases upward through levels A-D to whatever maximum value exists at level D. For the purpose of modelling strain in the attachment zone, both the horizontal and vertical gradients of horizontal shear have been made linear across the zone. This is a reasonable assumption for this first-order approach. Non-linearity in the distribution of the shear components would result necessarily from a non-uniform thickness of attachment zone and/or from the differences in rheological behaviour within the attachment

105

zone. Nevertheless, the solutions presented here for an ideal attachment zone of uniform thickness should capture the major features of attachmentrelated strain. Foliation and lineation patterns Figure 4 shows a series of strip maps representing levels A-D of an <2/8-thick attachment zone, above the wrench zone, with wrench shear y(w) — 1 and y(w) = 3. The maps display the orientation of foliation and lineation, taken to be the XY plane and the X direction, respectively, of the finite strain ellipsoid. These maps represent half of the attachment zone (Fig. 3), from the block centre to the edge, for a dextral wrench system. In the other half of the attachment zone, foliation planes have the same strikes but opposite dips, and lineations plunge in the opposite direction, owing to the change in shear sense of the horizontal shear component across the centre of the zone. Figure 5 shows two cross-sections through the attachment zone, for a wrench shear y(w) = 1 and y(w) = 3, and the associated graph of strain gradients. An attachment zone 2.5 km thick is shown as an example, for a block that would be 20 km wide. However, the results shown are, in principle, non-dimensional because the strain is a function of the ratio between attachment thickness and block width. The central part of the attachment zone has the geometry of a funnelshaped syncline; at the centre of the attachment zone, foliation is vertical, and at the edge, foliation dips only 20-25°. Therefore, for an attachment beneath an array of rigid blocks with intervening strike-slip faults, the modelling predicts an alternation between funnel-shaped synforms beneath block centres, and archshaped antiforms beneath the strike-slip faults (Teyssier et al. 2002). Lineation always plunges shallowly (stereonets, Fig. 4), due to the wrench component near the centre of the attachment and also to the increasingly large horizontal shear strain toward the edge of the attachment zone. Strain gradients Finite strain varies dramatically through the model attachment zone (Fig. 5). Strains are up to X/Z > 20 for wrench shear of y(w) — 1 and X/Z > 150 for wrench shear of y(w) = 3 in the upper corners of the attachment zones beneath the edges of rigid blocks. Similarly, strain gradients are also developed vertically through the attachment zone (Fig. 5). Figure 6 shows vertical strain gradient profiles at various distances from the centre of attachment zones for a 20 km wide

106

C. TEYSSIER & L. CRUZ

Fig. 4. Strip maps and stereoplots showing lateral and vertical variations of foliation and lineation from level A to level D in a wrench attachment zone (only half of solutions shown); orientations are consistent with rigid blocks being displaced by north-south strike-slip faults. Model results are shown for wrench shear strain of 1 (top) and 3 (bottom). Attachment zones show relatively consistent patterns of foliation with a shallowing of dip toward the top and beneath the rigid blocks' edges; in this region, lineations are also rotated into parallelism with the horizontal shear direction. In an attachment zone, the foliation trajectories are reminiscent of a folded pattern, as shown in the stereoplots.

rigid block. Figure 7 illustrates graphically the vertical and horizontal gradients in the ellipticity of the XZ finite strain ellipse. Next we investigate the effect of a wrench shear gradient that could permeate downward from the strike-slip fault and perturb the first-order strain patterns in this region of the attachment zone. This shear gradient is similar to that associated with a mode III crack (Pollard & Segall 1987), with maximum shear at the top of the attachment zone, decreasing to the background shear of T(W) — 1 at the bottom. In this case, the shear gradient is distributed over a zone of finite width equivalent to a mode III (or scissor) shear zone. Over what width should that deformation be distributed? It is reasonable to consider that the width of attachment zone affected by the wrench shear gradient, herein called the 'width of influence,' scales with the thickness of the attachment zone. This assumption is based on the same mechanical arguments that were used to define

an attachment zone in the first place, and also on the analogy with the typical width of influence along locked strike-slip faults (Savage & Burford 1973). Consider the 2.5 km thick attachment zone described in Figure 5, with the distribution of wrench shear y(w) and horizontal shear y(ns) as shown on the grid in Figure 3. The modification to this pattern due to the strike-slip fault 'width of influence' is shown in Figure 8, and spans a region ~2.5 km wide on either side of the strike-slip fault. A motion of 20km on such a fault, corresponding to a wrench shear y(w) = 1, imparts an additional y(w) = 4 over me 5 km wide zone in level D. This wrench shear decreases downward to the background shear strain of 7(w) = 1 in level A, and also decreases laterally to match the wrench shear values beyond the width of influence (Fig. 8a). Strain was calculated and foliation orientations were determined involving these possible influences. Figure 8b and c

STRAIN GRADIENTS IN ATTACHMENT ZONES

107

Fig. 5. Strain gradients in wrench attachment zones beneath rigid blocks, for wrench shear strains of 1 (top) and 3 (bottom). Cross-sections show the patterns of modelled foliation (tick marks) and contours of the finite strain gradient, also shown in graph form across the attachment zone, for levels A-D. Attachment zones are characterized by a funnel-shaped synformal structure beneath the centre of rigid blocks, with foliation shallowing toward the block margins. Very large finite strain gradients develop laterally in the upper level and vertically beneath the edges of the rigid blocks.

show the model foliation trajectories on a crosssection of the modified attachment zone. The basic foliation pattern is similar to that for the simpler case (Figs 4 & 5), but the modifications cause the foliation to steepen beneath the strikeslip faults, creating a prominent inflection in the foliation trajectories (Fig. 8c).

Transpression and transtension attachments This section now explores attachment zones with kinematic boundary conditions that deviate from

pure wrenching. We consider the ductile crust at depth deforming relatively homogeneously in transpression and transtension. We reason again that, if upper crustal blocks behave rigidly, attachment zones should develop beneath them. For the calculations presented here, and because we consider only low angles of convergence or divergence, the specific style of contraction or extension in the upper crustal blocks is simplified. In principle, the displacement trajectories imposed by such details (i.e. the number and distribution of faults accommodating contraction or extension) would affect the results only as a second-order

108

C. TEYSSIER & L. CRUZ

Fig. 6. Plots of vertical strain gradients across levels A-D of attachment zones for wrench shear strains of 1 (top) and 3 (bottom), and for two distinct ratios of block width/attachment thickness. Vertical gradients are shown at various locations of the attachment zone, normalized to block width a. For example, for block width a — 20 km, gradients are shown for vertical profiles located at <2/8 (2.5 km), a/4 (5 km), 3#/8 (7.5 km) and a/2 (10 km) from the centre of the block, as shown.

phenomenon - see for example the concept of orogenic float in Oldow et al (1990). The angle of convergence or divergence is defined by the ratio of contraction or extension across the deformation zone over the wrench component of displacement. In the following cases, we treat only low angles of convergence and divergence, with y= 1 in all cases. These cases correspond to 10%, 25% and 50% contraction or extension for angles of convergence or divergence of approximately 5.7°, 14° and 26.5°, respectively (Fig. 9).

Transpression attachments In transpression attachments, foliation patterns are similar to those developed in pure wrenching (Figs 4, 6 & 8), except that, in general, the foliation is more steeply dipping; the added pure shear component tends to steepen the XZ plane of the finite strain ellipsoid. For the case of 50% contraction shown in Figure 9, the X-axis of the finite strain ellipsoid is vertical in the transpression zone below the attachment (Fig. 10). It

has been previously established that the X-axis is always vertical in transpression zones if the angle of convergence is over 20° (Tikoff & Teyssier 1994). For convergence angles less than 20° the X-axis begins horizontal then switches to vertical at critical values as the finite strain ellipsoid goes through a pure flattening shape (Tikoff & Greene 1997; Teyssier & Tikoff 1999). In the other cases studied here for 10% and 25% contraction and y(w) = 1, the finite strain is not sufficiently large for this horizontal-to-vertical switch to occur. In the case shown for 50% contraction, the lineations in the attachment zone are shallowly plunging, especially at the edges of the attachment zone, because of the strong effect of the horizontal shear component (Fig. 10). As previously predicted for transpressional systems, finite strain ellipsoids develop in the flattening field (Sanderson & Marchini 1984; Fossen & Tikoff 1993). Figure 10 shows a Flinn diagram with the shape of the finite strain ellipsoid in the transpression zone marked as W, well in the flattening field for the three

STRAIN GRADIENTS IN ATTACHMENT ZONES

109

is located at the same place. This means that the gradients in the shape of the finite strain ellipsoid are the same, irrespective of the angle of convergence and even independent of the amount of strain. Therefore, this model strain gradient is a very robust feature of transpressional attachment zones, and should be observed if these types of attachment zones indeed develop in Nature.

Fig. 7. Visual representation of the shape and orientation of the finite strain ellipsoid in a dextral wrench attachment zone developed after a shear strain of 1 (top) and 3 (bottom) illustrating vertical and horizontal strain gradients from centre to edge of blocks.

cases of oblique convergence considered here. However, the superposition of an increasingly large component of horizontal shear through the attachment zone, outward and upward, takes the finite strain through plane strain and even into severe constriction. The points in the Flinn diagram marked A, B, C and D (Fig. 10) correspond to various regions of the attachment zone, as shown in the inset, for different degrees of contraction across the zone. Although homogeneous transpression develops flattening strain, a strong constriction fabric forms in the upper parts of transpressional attachment zones. This constriction arises because the horizontal shear component that shortens rocks in the near-vertical direction, especially at high strain, combines with the pure shear component that shortens rocks in the horizontal direction (arrows on the cross-sections shown in Fig. 10). The cross-sections across transpression attachments for 10% and 25% contraction demonstrate that the surface within the attachment zone where plane strain is expected to separate the two regions with flattening below and constriction above

Transtension attachments In transtension, the model strain results are perfectly symmetrical with transpression. Figure 11 illustrates the main results for transtensional fabric orientation and shapes of the finite strain ellipsoids. The cross-section for 50% extension shows a flat-lying foliation in the transtension zone below, as well as in the centre of the attachment zone (marked W in the inset). For divergence angles greater than 20°, foliation is always horizontal (Teyssier & Tikoff 1999), and for divergence angles lesser than 20°, foliation forms vertical and may switch to horizontal, as the finite strain ellipsoid goes through pure constriction. For two of the examples shown here (10% and 25% extension), finite strain is not large enough for this switch to occur. In our model transtension attachment zones, foliation is more shallowly dipping than in the transpression cases because the pure shear component of transtension tends to flatten the foliation. The shape of the finite strain ellipsoid developed in our model transtension zones changes from deep constriction, as previous work has shown (Teyssier & Tikoff 1999), to highly flattened upward and outward, as shown in Figure 11. This flattening in the attachment is explained by the combined effect of the pure shear component, that tends to elongate rocks in a direction normal to the deformation zone (see arrows on cross-sections in Fig. 11), and the horizontal shear, that elongates rocks in the direction of the strike-slip fault (north-south in Fig. 11). These two directions of extension combine to produce flattening fabrics. As for the transpression attachment case above, the location of the transition between constriction and flattening is extremely robust and is once again independent of finite strain and angle of divergence. Such boundaries and gradients in the shape of the finite strain ellipsoid should be characteristic of fossil attachment zones developed in transtension. Summary If attachment zones develop in transpression and transtension settings, our models predict a pattern of foliation/lineation similar to that

110

C. TEYSSIER & L. CRUZ

Fig. 8. Cross-section of a wrench attachment zone in which the wrench shear y(w) shows a gradient beneath the strike-slip fault, (a) Grid for strain calculation in the 'width of influence'; the 'width of influence' over which the wrench shear gradient exists is equal to the thickness of the attachment zone, (b) Segment of foliation trace calculated at discrete points in attachment zone, (c) Foliation trajectories in attachment zone, showing an inflection due to the wrench shear gradient beneath the strike-slip fault.

produced in wrench zones. The major difference lies in the distribution of flattening and constrictional finite strains in the attachment zones. For a given attachment system, the location of the boundary between flattening and constriction

domains (the plane strain surface shown in Figs 10 & 11) is a marker characteristic of this type of attachment. In the cases shown in Figures 10 and 11, where y(w) = 1, the plane strain surface separating the flattening and constriction

Fig. 9. Cases of transpression and transtension used in this chapter to model strain in attachment zones developed during low-angle oblique convergence and divergence.

STRAIN GRADIENTS IN ATTACHMENT ZONES

111

Fig. 10. Flinn diagram, orientation data and cross-section of attachment zone developed in transpression attachment zone after a finite displacement produced by y(w) — 1 and 10%, 25% and 50% shortening. In the transpression zone below the attachment zone, as well as on a vertical column beneath the centre of rigid blocks (marked W), finite strain is flattening, as expected from transpression; however, moving toward the edge and the top of the attachment zone, finite strain becomes plane strain and even severely constrictional. The position of the plane strain surface (on cross-section) is stationary whatever the amount of finite strain or angle of convergence (here shown for 10% and 25% shortening), making it a robust criterion. Note that the orientation of lineation in the transpression zone below the attachment zone (W in stereoplot) switches from horizontal to vertical as the angle of convergence increases and finite strain goes through pure flattening; however, lineation in the attachment zone is shallowly plunging, due to the increased effect of the horizontal shear component.

regions is located where y(hs) = 1. If y(w) = n, then the plane strain surface would be located where y(hs) = n. However, the plane strain surface would still be at the same location, because the horizontal shear strain scales to the wrench shear strain; for very high strains large y(w) and y(hs) - this surface would separate domains of more severe flattening and constriction, but would remain at the same depth in the attachment zone. Therefore, the position of the plane strain envelope is a robust result that can only be altered by changing the horizontal and vertical gradients in horizontal shear strain imposed in our model attachment zones. In the examples studied here, the vertical and horizontal

gradients of horizontal shear are linear; were they non-linear, the position of the plane strain surface would change, but an analogous fabric boundary would still exist. Figure 12a displays the finite strain X/Z gradients that develop in the various transpression and transtension attachments we considered above (Figs 10 & 11). The model gradients are similar, regardless of the amount of contraction or extension across the deformation zone, because y(hs) dominates the finite strain; this component is the same whatever the amount of contraction or extension across the zone. The finite kinematic vorticity also shows significant vertical and horizontal gradients across the model attachments (Fig. 12b). The base and centres

112

C. TEYSSIER & L. CRUZ

Fig. 11. Flinn diagram, orientation data and cross-section of attachment zone developed in transtension attachment zone after a shear strain y(w) = 1 and 10%, 25% and 50% extension. In the transtension zone below the attachment, as well as on a vertical column beneath the centre of rigid blocks (marked W), finite strain is constriction; however, moving toward the edge and the top of the attachment zone, finite strain becomes plane strain and even severely flattening. The position of the plane strain surface (on cross-section) is stationary whatever the amount of finite strain or the angle of divergence. Note that the orientation of foliation in the transtension zone below the attachment zone switches from vertical to horizontal as the angle of divergence increases and finite strain goes through pure constriction.

of attachment zones are characterized by a larger component of coaxial strain, while the outer edges are dominated by strong non-coaxial flow, associated with horizontal simple shearing.

Discussion Attachment zones are predicted to develop beneath rigid blocks that undergo translation

STRAIN GRADIENTS IN ATTACHMENT ZONES

113

Fig. 12. (a) Finite strain gradients (shown as X/Z ratio) at the edge of transpression and transtension attachment zones, across levels A through D, for y(w) = 1. Gradients are similar for various degrees of transpression and transtension (10%, 25%, 50%). (b) Lateral (top) and vertical (bottom) gradients in kinematic vorticity in transpression or transtension zones characterized by 10%, 25% and 50% shortening/extension (defined in Fig. 9). The coaxial component of flow decreases toward the top and the edges of attachment zones due to the increased horizontal simple shear in these regions.

along strike-slip faults. Several lines of evidence suggest that these attachments do in fact exist in Nature and at several different scales. In wrench settings, subhorizontal reflectors have been observed at the brittle-ductile transition (Jones et al. 1994). Such reflectors occur in a zone of finite thickness (several kilometres) that can be interpreted as an accommodation zone between the upper and lower crust, which move differently. Garde et al. (2002) recently interpreted exhumed structures in Greenland as related to a partitioned and attached transpression system; Carreras & Capela (1995) reported similar zones in the Pyrenees that can be interpreted as attachments. In addition, Tikoff et al. (2002) have argued that the lower crust (i.e. granulite terrains), which is commonly characterized by subhorizontal fabrics and very large shear strains, may have 'attached' the crust and the mantle while these two layers deformed differently. Therefore, granulite terrains could indeed represent large domains of crust formed by attachment tectonics. In broad continental deformation zones such as the Tibetan orogen, large segments of lithosphere have moved by wrench or transpression motion (Tapponnier et al. 2001).

It is conceivable the large seismic anisotropy that generally parallels major, lithospheric-scale strikeslip faults, like the Kun-Lun fault (Herquel et al. 1999), the San Andreas fault (Ozalaybey & Savage 1995), or the El Pilar-Central Range faults in the Caribbean-South American plate boundary zone (Russo et al. 1996), result from attachment zone fabrics formed at the base of the lithosphere. For lithospheric blocks several hundred kilometres wide, attachment zones could be up to 50-100 km thick. Our modelling suggests that the large strain generated by the strong horizontal shear component could result in strong olivine fabric and associated seismic anisotropy.

Conclusions In wrench, transpression, or transtension settings, at major rheological boundaries within the crust or lithosphere, the potential exists for attachment zones to develop. The main results of the modelling presented here are summarized below: 1

The horizontal shear that is generated by sliding of rigid blocks over the top of more homogeneously deforming rocks beneath

114

2

3

4

C TEYSSIER & L. CRUZ could cause large strain gradients in attachment zones. It is this horizontal shear that dominates strain and fabric development in the model attachment zones. Beneath translating rigid upper crustal blocks, funnel-shaped synforms of foliation develop as a result of attachment tectonics. With increasing strain in attachment zones, lineations become parallel to the upper bounding strike-slip faults. Sense of shear reverses across the axial surfaces of the synforms. Beneath the strike-slip faults that separate upper crustal blocks, foliation is predicted to steepen in a zone whose width may scale to the attachment zone thickness (Fig. 8c). In transpression and transtension attachments with angles of convergence or divergence less than 20°, the pattern of foliation and lineation is similar to the case of pure wrenching. In transpression attachment zones, the fabrics change drastically from flattening at the base to constriction near the top. Similarly, in transtension attachments, fabrics change from constriction at the base grading to flattening upward. The zones of flattening and constriction are separated by a plane strain surface whose position is independent of amount of strain or obliquity of motion and should be a robust signature of transpression and transtension attachment zones.

This research was initiated with support from NSF-EAR 9607018 and Grant-in-Aid 17904 from the University of Minnesota Graduate School, and completed with support from NSF-EAR 0126166. We are grateful to Basil Tikoff and the Minnesota Structure and Tectonics Group for stimulating discussions, John Weber for enhancing the clarity of this paper, Michel de Saint Blanquat for his constructive review and for adding some Cartesian logic to the text, John Grocott for his encouragements and editorial work, and Dazhi Jiang and Ray Fletcher for pointing out some important missing points.

References BOURNE, S.J., ENGLAND, P.C. & PARSONS, B. 1998a. The motion of crustal blocks driven by flow of the lower lithosphere and implications for slip rates of continental strike-slip faults. Nature, 391, 655-659. BOURNE, S.J., ARNADOTTIR, T., et al 1998£. Crustal deformation of the Marlborough fault zone in the South Island of New Zealand: geodetic constraints over the interval 1982-1994. Journal of Geophysical Research, 103, 30147-30165. CARRERAS, J. & CAPELA, I. 1995. Tectonic levels in the Paleozoic basement of the Pyrenees: a review and new interpretation: Reply. Journal of Structural Geology, 17, 1493-1495.

ENGLAND, P.G. & WELLS, R.E. 1991. Neogene rotations and quasicontinuous deformation of the Pacific Northwest continental margin. Geology, 19, 978-981. FOSSEN, H. & TIKOFF, B. 1993. The deformation matrix for simultaneous simple shearing, pure shearing, and volume change, and its application to transpression/transtension tectonics. Journal of Structural Geology, 15, 413-422. GARDE, A., CHADWICK, B., GROCOTT, J., HAMILTON, M.A., MCCAFFREY, K.J.W. & SWAGER, C.P. 2002. Mid-crustal partitioning and attachment during oblique convergence in an arc system, Palaeoproterozoic Ketilidian Orogen, southern Greenland. Journal of the Geological Society of London, 159, 247-261. HERQUEL, G., TAPPONNIER, P., WITTLINGER, G., MEI, J. & DANIAN, S. 1999. Teleseismic shearwave splitting and lithospheric anisotropy beneath and across the Altyn Tagh Fault. Geophysical Research Letters, 26, 3225-3228. JONES, D.L., GRAYMER, R., WANG, C., MC£VILLY, TV. & LOMAX, A. 1994. Neogene transpressive evolution of the California Coast Ranges. Tectonics, 13, 561-574. KOHLSTEDT, D.L., EVANS, B. & MACKWELL, S.J.

1995. Strength of the lithosphere: constraints imposed by laboratory experiments. Journal of Geophysical Research, 100, 17587-17602. MERLE, O. & GAPAIS, D. 1997. Strains in thrustwrench zones. Journal of Structural Geology, 19, 1011-1014. MOLNAR, P. 1988. Continental tectonics in the aftermath of plate tectonics. Nature, 335, 131-137. MOLNAR, P. 1992. Brace-Goetze strength profiles, the partitioning of strike-slip and thrust faulting at zones of oblique convergence, and the stressheat flow paradox of the San Andreas fault. In: EVANS, B. & WONG, T.-F. (eds) Fault Mechanics and Transport Properties of Rocks. Academic Press, New York, 435-459. OLDOW, J.S., BALLY, A.W. & AVE-LALLEMANT, H.G. 1990. Transpression, orogenic float, and lithospheric balance. Geology 18, 991-994. OZALAYBEY, S. & SAVAGE, M.K. 1995. Shear-wave splitting beneath western United States in relation to plate tectonics. Journal of Geophysical Research, 100, 18135-18149. POLLARD, D.D. & SEGALL, P. 1987. Theoretical displacements and stresses near fractures in rock: with applications to faults, joints, veins, dikes, and solution surfaces. In: ATKINSON, B.K. (ed.) Fracture Mechanics of Rock., Academic Press, New York, 277-349. Russo, R.M., SILVER, P.G., FRANKE, M., AMBEH, W.B. & JAMES, D.E. 1996. Shear-wave splitting in northeast Venezuela, Trinidad, and the eastern Caribbean. Physics of the Earth and Planetary Interiors, 95, 251-275. SANDERSON, D. & MARCHINI, R.D. 1984. Transpression. Journal of Structural Geology, 6, 449-458. SAVAGE, J.C. & BURFORD, R.O. 1973. Geodetic determination of relative plate motion in central

STRAIN GRADIENTS IN ATTACHMENT ZONES California. Journal of Geophysical Research, 78, 832-845. SAVAGE, J.C., SVARC, J.L. & PRESCOTT, W.H. 1999. Geodetic estimates of fault slip rates in the San Francisco Bay area. Journal of Geophysical Research, 104, 4995-5002. SYLVESTER, A.G. 1988. Strike-slip faults. Geological Society of America Bulletin, 100, 1666-1703. TAPPONNIER, P., Xu, Z., ROGER, R, MEYER, B., ARNAUD, N., WITTLINGER, G. & YANG, J. 2001. Oblique stepwise rise and growth of the Tibet Plateau. Science, 294, 1671-1677. TEYSSIER, C. & TIKOFF, B. 1999. Fabric stability in oblique convergence and divergence. Journal of Structural Geology, 21, 969-974. TEYSSIER, C., TIKOFF, B. & WEBER, J. 2002. Attachment between brittle and ductile crust at wrenching plate boundaries. European Geophysical Union Stephan Mueller Special Publication Series, 1, 75-91. TIKOFF, B. & FOSSEN, H. 1993. Simultaneous pure and simple shear: the unified deformation matrix. Tectonophysics, 217, 267-283. TIKOFF, B. & FOSSEN, H. 1996. Computer applications for visualization and calculation of deformation.

115

In: De PAOR, D. (ed.) Microcomputers and Structural Geology. Pergamon, Oxford, 75-96. TIKOFF, B. & GREENE, D. 1997. Stretching lineations in transpressional shear zones. Journal of Structural Geology, 19, 29-40. TIKOFF, B. & TEYSSIER, C. 1994. Strain modeling of displacement-field partitioning in transpressional orogens. Journal of Structural Geology, 16, 1575-1588. TIKOFF, B., TEYSSIER, C. & WATERS, C. 2002. Clutch tectonics and the partial attachment of lithospheric layers. European Geophysical Union Stephan Mueller Special Publication Series, 1, 57—73. VAUCHEZ, A. & TOMMASI, A. 2003. Wrench faults down to the asthenosphere: geological and geophysical evidence and thermo-mechanical effects. In: STORTI, F., HOLDSWORTH, R.E. & SALVINI, F. (eds)Intraplate Strike-Slip Deformation Belts., Geological Society, London, Special Publications, 210, 15-34. VAUCHEZ, A., TOMMASI, A. & BARRUOL, G. 1998. Rheological heterogeneity, mechanical anisotropy, and tectonics of the continental lithosphere. Tectonophysics, 296, 61-86.

This page intentionally left blank

The role of crustal heterogeneity in controlling vertical coupling during Laramide shortening and the development of the Caribbean-North America transform boundary in southern Mexico: insights from analogue models MARIANO CERCA1, LUCA FERRARI1, MARCO BONINI2, GIACOMO CORTI2'3 & PIERO MANETTI4 1 Centra de Geociencias, Universidad National Autonoma de Mexico, Campus Juriquilla, Apartado Postal 1-742, Queretaro 76230, Mexico (e-mail: mcerca @ geociencias. unam. mx, luca @ geociencias. unam. mx) 2 Consiglio Nazionale delle Ricerche, Istituto di Geoscienze e Georisorse, Sezione di Firenze, via G. La Pira 4, 50121 Firenze, Italy 3 Dipartimento di Scienze della Terra, Universita degli Studi di Pisa, via S. Maria 53, 56126 Pisa, Italy Consiglio Nazionale delle Ricerche, Istituto di Geoscienze e Georisorse, via G. Moruzzi 1, 56124 Pisa, Italy Abstract: Analogue models of polyphase deformation involving crustal differences in strength, thickness and density give insights into lateral and vertical strain propagation during Late Cretaceous shortening and Early Tertiary left-lateral shearing related to the early development of the North America-Caribbean plate boundary in southern Mexico. Analogue models reproduce a two-phase deformation characterized by a first stage of compression orthogonal to the plate boundary, simulating deformation induced by the Laramide orogeny, followed by a later stage of left-lateral transpression associated with the transfer of the Chortis block from the North American to the Caribbean plate during the early stage of development of the new plate boundary in Early Tertiary times. Based on detailed structural observations in the Guerrero-Morelos platform and the western part of the Mixteco terrane of southern Mexico, we document that a transpressive regime affected continental red bed sequences of Early Paleocene to Late Eocene, and rotated and refolded Laramide structures during this second phase. Our model ends before the transtensional regime that affected the region, which is marked by a volcanic episode of Late Eocene-Oligocene. This change in the deformation regime records the passage of the NW tip of the Chortis block (North America-Cocos-Caribbean triple junction), when subduction replaced transform faulting along the southern Mexico margin. The models focus on the structures formed around the flanks of a thicker/more rigid crustal block that simulates the rock assemblages of the Palaeozoic orogens of southern Mexico (Mixteco-Oaxaca-Juarez block, MOJB). The comparison of the mechanism of deformation of three different analogue models with the natural prototype explains most of the structures observed around the MOJB. Counterclockwise vertical-axis rotations of pre-existing structures in the western flank of the MOJB observed in the GuerreroMorelos platform are consistent with the modelled structures. Vertical movements of the modelled MOJB induced by the transpressive regime can explain the Papalutla thrust and the basement upheaval and gravitational sliding of the cover in the Tentzo Ranges observed at the western and northern margins of the MOJB, respectively. The growth and propagation of thrusting controlled by the geometry of the block along the eastern margin also correlates with the Vista Hermosa fault. The propagation of strain to the north increases with higher contrast in strength of the thick block with respect to the adjacent modelled crust. Analogue modelling failed to reproduce all the structural details of southern Mexico and, specifically, the structures observed inside the MOJB. The latter, however, are controlled by pre-existing discontinuities, which are not simulated in the model. As a whole, the results demonstrate that crustal heterogeneity in a developing left-lateral plate boundary zone produces a stronger vertical coupling between ductile and brittle crust and a widening of the deformation zone along the margin of the North America plate in southern Mexico. From: GROCOTT, J., MCCAFFREY, K. J. W., TAYLOR, G. & TIKOFF, B. (eds) 2004. Vertical Coupling and Decoupling in the Lithosphere, Geological Society, London, Special Publications, 227, 117-140. 0305-8719/047$ 15 © The Geological Society of London 2004.

118

M. CERCACTAL.

Unlike convergent margins, deformation along transform plate boundaries on continental lithosphere can affect a relatively narrow zone on both sides of the plates (e.g. North Anatolia fault, Polochic-Motagua fault system). However, it can be argued that, during the initial development of a transform boundary, deformation is most commonly accommodated in a wide zone, whereas in a mature stage strain localizes along discrete systems of lithospheric strike-slip faults (e.g. Gordon 1998). The degree of coupling between the upper mantle, the lower crust and the upper crust ultimately controls the width of the deformation zone at the early stage of development of the boundary. Furthermore, the presence of crustal blocks with different thicknesses and strengths is likely to alter the coupling and to control the transmission of deformation toward more internal zones of the plates. The development of the North AmericaCaribbean plate boundary in the Early Tertiary can help to elucidate the importance of crustal heterogeneity during the early development of a transform boundary. It has been suggested that the Chortis block of Central America was an integral part of the North America plate from Jurassic to Late Cretaceous times (Meschede &

Frisch 1998, and references therein). At the end of Late Cretaceous, the Chortis block began to detach from North America and to move eastward with the Caribbean plate, likely as a consequence of a reorientation from normal to oblique subduction of the Farallon plate (Herrmann et al. 1994; Meschede et al. 1996). As a result, the continental margin of southern Mexico was truncated (Riller et al. 1992; Herrmann et al. 1994; Schaaf et al. 1995) and middle to lower crustal rocks (Xolapa complex) were exhumed along a ~60 km wide band to the north of new plate boundary (Moran-Zenteno et al. 1996). Non-coaxial deformation and migmatization were inferred to have developed in the Xolapa complex between 70 and 46 Ma (Herrmann et al. 1994; Meschede et al. 1996) or in Early Cretaceous (Moran-Zenteno 1992; Ducea et al. 2003). Mylonitic zones developed in a general left-lateral transtensional regime are observed bounding middle to lower crustal rocks of the Xolapa complex to the north (Fig. 1). Vertical propagation of strain in the upper crust and horizontal transmission to the north of the shear zone that bounds the Xolapa complex are not well understood. Meschede et al. (1996) used inversion of brittle microstructures

Fig. 1. Terrane boundaries (thick dashed grey lines) and major structural features of southern Mexico (modified after Campa & Coney 1983; Sedlock et al. 1993). TMVB, trans-Mexican volcanic belt. Terranes: G, Guerrero; M, Mixteco; O, Oaxaca; J, Juarez; Ma, Maya; and X, Xolapa. GMP, Cretaceous Guerrero-Morelos platform. Structures: (1) Teloloapan-Ixtapan de la Sal thrust; (2) Zitlala-Cuernavaca thrust; (3) refolded and rotated Laramide folds; (4) Papalutla thrust; (5) Tentzo Ranges arcuate folds; (6) Oaxaca fault; and (7) Vista Hermosa thrust. Along the northern boundary of the Xolapa terrane, several mylonitic zones (Mz) crop out in the Tierra Colorada shear zone (TC), and Chacalapa fault (Ch). Black dashed lines show the approximate boundaries of the Mixteco-Oaxaca- Juarez block (MOJB) and black solid lines mark thrust boundaries.

COUPLING DURING LARAMIDE SHORTENING

measured in a wide area of southern Mexico to claim that the stress applied at the plate boundary has been transmitted to the north of the Xolapa complex into the Mixteco and Oaxaca terranes (Fig. 2). They grouped structures in Palaeozoic to Early Tertiary rocks inferred to have developed between 70 and 40 Ma in a unique event characterized by subhorizontal (TI and cr3, i.e. a left-lateral strike-slip regime of deformation. Recently, however, we have documented a more complex deformation history. Detailed field studies in the Guerrero-Morelos platform and the eastern part of the Mixteco terrane (Fig. 1) show that a major episode of east-west

119

shortening between ~88 and 67 Ma (Laramide orogeny) was followed by Early Tertiary leftlateral transpression that affected an area up to 250km to the north of the modern plate boundary (Cerca & Ferrari 2001). Starting from Latest Eocene, transpression was replaced by transtension, which triggered widespread silicic volcanism (Moran-Zenteno et al. 1999; AlanizAlvarez et al. 2002). The Early Tertiary transpression was particularly diffuse at the boundary and in the eastern part of the Mixteco-Oaxaca Juarez block (MOJB), a thicker and more rigid crustal block, suggesting that it controlled the widening of the deformation zone related to the

Fig. 2. Sketch map showing the main lithological units of southern Mexico (modified after Ortega-Gutierrez et al. 1992; Consejo de Recursos Minerales 2001). GC, Guichicovi complex. Solid lines indicate major fault zones.

120

M. CERCAErAL.

development of the Caribbean-North America plate boundary. In this Chapter we approach the study of deformational features around the MOJB indirectly, by performing analogue models designed to investigate the space-time propagation of deformation in relation to crustal rheological heterogeneities during polyphase deformation simulating the tectonic evolution of the southern Mexico deformed margin.

Geological and tectonic setting Crustal structure of southern Mexico and the Mixteco—Oaxaca—Juarez block Geologically, Mexico south of the trans-Mexican volcanic belt (TMVB) consists of a heterogeneous mosaic of crustal blocks (Fig. 1) that have been traditionally classified using the tectonostratigraphic terrane analysis (Campa & Coney 1983; Sedlock et al 1993). Owing to their differing geological histories, these crustal blocks or terranes have distinct thicknesses and rheologies that must be considered in any deformation modelling. The Mixteco and Oaxaca terranes are considered to have the oldest basement in southern Mexico. These terranes are mainly composed of Precambrian or Palaeozoic metamorphic rocks, and a Jurassic to Early Tertiary sedimentary and volcanic cover (Campa & Coney 1983; Sedlock et al 1993). The Mixteco terrane records a Late OrdovicianEarly Silurian continental collisional event (Acatecan orogeny) related to the closure of the lapetus Ocean (Ortega-Gutierrez et al. 1999) and was later sutured to the Grenvillian Oaxaca terrane in the Early Permian (Elfas-Herrera & Ortega-Gutierrez 2002). The basement of the Juarez terrane is poorly known, but recent studies suggest that it could be also pre-Mesozoic. The boundary between the Oaxaca terrane and the Juarez terrane is the north-south-trending Sierra de Juarez mylonitic complex, which records right-lateral movements related to the southward movement of the Yucatan block and the opening of the Gulf of Mexico in Mid-Jurassic times (Alaniz-Alvarez et al 1996) (Fig. 1). It has also been suggested that the Juarez terrane was the site of a rifting in Jurassic times (Sedlock et al 1993, and references therein). However, Jurassic volcanism is very limited and recently published regional maps show a wide area of Palaeozoic metamorphic rocks in the core of the Juarez terrane (Consejo de Recursos Minerales 1998, 2001). In addition, the protolith of the Sierra de Juarez mylonites is, at least in part, the Grenvillian

Oaxaca terrane (Alaniz-Alvarez et al 1996), and Grenvillian rocks have been reported in the Guichicovi complex SW of the Juarez terrane (Murillo et al 1992; Weber & Kohler 1999) (Fig. 2), which suggests a lateral continuity of the rocks in between. This seems to be confirmed by a magnetotelluric study crossing the Oaxaca and Juarez terranes, which indicates that they could share a similar basement at depth (Jording et al 2000). The eastern boundary of the Juarez terrane is the Vista Hermosa fault zone (Ortega-Gutierrez et al 1990), east of which is the transitional crust of the Maya terrane thinned during the opening of the Gulf of Mexico (Fig. 1). Given its geological characteristics, we consider that the block (MOJB) formed by the Mixteco, Oaxaca and Juarez terranes presents a more rigid basement and a thicker crust than the surrounding regions. To the west of this block, Mesozoic island arc assemblages and Cretaceous marine carbonates crop out extensively (Guerrero terrane and Guerrero-Morelos platform, Figs 1 & 2) and the crust is relatively thinner. The western limit of the block corresponds to the contact between the deformed and metamorphosed basement (Acatlan complex) of the Mixteco terrane and the Cretaceous carbonate sequences of the Guerrero-Morelos platform. Finally, further geometrical constraints are provided by the presence of a semiarc-shaped fold-and-thrust belt that surrounds and follows approximately the border of the MOJB (Figs 1 & 2). Available geophysical data indicate that the crustal thickness decreases from 45 km in the central part of the MOJB to 28 km north of Zihuatanejo and to 25 km in the Tehuantepec Isthmus zone (Campos-Enriquez & SanchezZamora 2000; Valdez et al 1986; UrrutiaFucugauchi & Flores-Ruiz 1996; Garcia-Perez & Urrutia-Fucugauchi 1997). Thinning or absence of the Late Cretaceous carbonate sequences in the Mixteco, Oaxaca and Juarez terranes indicates that the MOJB was at least partially emergent and represented a major heterogeneity of the southern Mexico crust with well-developed boundaries by Late Cretaceous times (Fig. 3). Laramide deformation After earlier deformation phases, southern Mexico was affected in Late Cretaceous time by contractional deformation during the Laramide orogeny (Lang et al 1996; Bird 1998; Cabral-Cano et al 2000fl). The migration of deformation towards the continent has been associated to low-angle and high-velocity subduction in the western margin of

COUPLING DURING LARAMIDE SHORTENING

121

Fig. 3. (a) Digital elevation model of the southern Mexico continental margin and (b) idealized longitudinal section. Analogue models were designed to investigate the influence of a rigid block in the brittle crust. In the models the section was simplified assuming a stratified two-layer and uniformly thick crust.

the North America plate (Bunge & Grand 2000). The onset of the Laramide deformation in southern Mexico has been constrained by the deposition of the Mezcala flysch, which records a sudden change from carbonaceous to terrigenous sedimentation and represents the youngest deformed unit. This transition is set at the Cenomanian-Turonian boundary (c. 93 Ma) (Hernandez-Romano et al 1997) or at the Turonian-Coniacian boundary (c. 89 Ma) (Lang & Frerichs 1998). Almost all authors agree in placing the end of the Laramide episode in the Paleocene in view of time constraints north of the TMVB (Salinas-Prieto et al 2000, and references therein). However, volcanic and plutonic rocks of 67-62 Ma unaffected by Laramide-style deformation suggest a Maastrichtian age at least in the Guerrero-Morelos platform area (OrtegaGutierrez 1980; Meza-Figueroa et al 2001). The Laramide orogeny produced regional ENE-directed shortening that presumably amalgamated and stacked the volcanic arcs and sedimentary successions of the Guerrero terrane onto an attenuated continental crust (Cabral-Cano et al 20000, b). The result of the shortening is manifested in a wide north-south-striking fold-and-thrust belt with vergence towards the ENE. According to Salinas-Prieto et al. (2000),

progressive shortening caused a second ductile deformation with opposite vergence (backthrusting) of structures. Early Tertiary deformation The rock sequences recording the time interval between Maastrichtian and Late Eocene in the study area consist mainly of continental sedimentary deposits and minor volcanic rocks (e.g. Tetelcingo, Balsas, Oapan and other locally named formations) that fill basins bounded by north-south folds and thrusts formed during previous eastward shortening. Until recently, shortening structures of contrasting style affecting these Early Tertiary sequences have been attributed to the Laramide deformation. Complex patterns of shortening and associated strike-slip faults have been observed widely in these sequences north of the shear zone bounding the Xolapa complex. In particular, deformation decreasing gradually upwards can be observed in the continental deposits of the Balsas and Tetelcingo formations on the western flank of the MOJB (Fries 1960; De Cserna et al. 1980; Ortega-Gutierrez 1980) (Fig. 4). The most notable example is a wide deformation band

122

M. CERCACTAL.

Fig. 4. Simplified geology of the western margin of the MOJB (Mixteco terrane) and Guerrero-Morelos platform and location of the Early Tertiary basins deformed.

(~60km) in front of the NE-SW-trending Papalutla thrust that affects Tertiary volcanosedimentary deposits of the Copalillo and Tuzantlan basins (Fig. 4). Deformation, characterized by NW-directed shortening and NW-SE strikeslip faults, is more intense near the Papalutla fault (Cerca & Ferrari 2001). This structure dips to the east and along 9 km of its length thrusts Palaeozoic rocks of the Mixteco terrane on top of the Cretaceous sedimentary succession of the Guerrero-Morelos platform. Models that characterize Laramide deformation by eastverging thrusting of the Late Cretaceous sequences (Campa 1978; Campa & Coney 1983; Salinas-Prieto et al 2000) fail to explain the geometry of this fault. On the other hand, our recognition of a deformation consistent with the geometry of the Papalutla fault in the Copalillo and Tuzantlan continental deposits

(Figs 4 & 5) suggests that it moved during the Early Tertiary, although previous movements are not discarded. In the middle of the Guerrero-Morelos platform, where the thickness of the sequences in the Balsas basin reaches c. 500 m, deformation characterized by a large NW-SE fold and the absence of normal faults is worth nothing. The intensity of folding decreases towards the top of the sequence (Fig. 5a-c). Towards the southern part of the GuerreroMorelos platform, deformation is characterized by asymmetric NW-SE synclinorium-type structures (Chilpancingo basin) and NW-SE strike-slip faults (Fig. 4). Near the boundary with the Xolapa complex, the Late Cretaceous carbonates are refolded and aligned with eastwest-trending folds with vertical hinge lines (Fig. 5d). In this area, large outcrops of carbonate

COUPLING DURING LARAMIDE SHORTENING

123

Fig. 5. (a) Panoramic view to the NE of the Balsas sequence in the central Guerrero-Morelos platform showing a largescale antiform. (b) Variation in inclination of bedding at the core of the antiform shown in (a), from bottom to top Maastrichtian Tetelcingo (MT), Paleocene Balsas (PB), and Oapan (PO) formations, the sequence is covered by undeformed Oligocene volcanic rocks (OV). Growth strata denote syndepositional activity of this structure, (c) Deformation of a volcano-sedimentary sequence in the Tertiary basins in front of the Papalutla thrust, (d) View to the west in the Mexico-Acapulco highway showing Cretaceous limestone of Paleocene. (e) Vertical bedding of Paleocene conglomerate within a tight fold in the Yanhuitlan sequence.

breccias adjacent to the mylonites and underlying the Early Tertiary sequences are evidence of the rupture and detachment of the Chortis block (Mills 1998). Within the MOJB, localized folding associated with strike-slip movements of the same age has

been observed in the Yanhuitlan area (Fig. 5e). Other structures (mostly strike-slip faults and minor folds) within and east of the MOJB have suspected Paleocene to Eocene age and affect the plate margin in an area of variable width to the north of the shear zone bounding the Xolapa complex.

124

M. CERCAETAL

Tertiary deformation All the above evidence indicates that, in the Guerrero-Morelos platform and the western part of the MOJB, deformation during Early Tertiary times was distinct from the Late Cretaceous Laramide shortening and was essentially transpressional. Constraining Laramide shortening between the time interval from ~88 to ~67 Ma implies that in southern Mexico it commenced earlier and continued for a shorter time than in the north (75-40 Ma; Bird 1998, and references therein). This difference suggests that a change in the tectonic setting of southern Mexico occurred during the Late Maastrichtian- Early Paleocene interval. There is ample evidence that an ESE left-lateral strike-slip regime over a broad zone dominated the Cenozoic tectonics of the MOJB. This regime has been related to the detachment of the Chords block from North America and its transfer to the eastward-moving Caribbean plate. This process is likely to have begun at the end of the Late Cretaceous as a consequence of changes in the angles of convergence and subduction between the Farallon and North America plates (Engebretson et al. 1985; Ratschbacher et al. 1991; Herrmann et al 1994; Meschede et al 1996). A second important change occurred when the northwestern tip of the Chortis block and the trench-trench-transform triple junction passed along the coast of southern Mexico. According to the model of Moran-Zenteno et al (1996), the uplift of the continental margin and the exhumation of the middle crustal rocks of the Xolapa complex followed the passage of the triple junction. Uplift and exhumation were accomplished through the development of the mylonitic zone bounding the Xolapa complex to the north. Available isotopic ages and crosscutting relationships between plutons and the mylonitic zones also indicate that magmatism was active just before and after the triple junction passage (Schaaf et al 1995; Moran-Zentenoet al 1999). In the area of the Guerrero-Morelos platform, the triple junction passage is well documented by a widespread Late Eocene-Early Oligocene (36-30 Ma) volcanic episode (Moran-Zenteno et al 1999, and references therein; Cerca et al 2003). In the northern part of the GuerreroMorelos platform, this last volcanic episode has been associated with a transtensional regime (Alaniz-Alvarez et al 2002). Post-Eocene transcurrent and transtensional deformation is widespread also to the east. Indeed, Meschede et al (1996) obtained a strike-slip palaeotensor from faults in volcanic rocks at Chilapa (their site CHI2-S) that we dated at 32.7 Ma (Cerca et al 2003), and a transtensional palaeotensor by

inversion from faults in the Etla tuff (site ETVS2) dated at -17 Ma (Urrutia-Fucugauchi & Ferrusquia-Villafranca 2001). Summarizing all the above information, we propose that the Early Tertiary strain (65-36 Ma) constituted a phase of deformation different from the Late Cretaceous Laramide shortening and the post-Eocene transtension. This deformation is represented overall by localized tectonic dragging effects and small counterclockwise rotations about the vertical axis of previously formed structures and semi-rigid crustal blocks. Consistent counterclockwise rotation of Laramide structures and Early Tertiary sequences (Balsas formation) has also been found in palaeomagnetic studies (Molina-Garza et al 2003, and references therein). We propose that a transpressive regime also affected the study region during these times based on the following considerations: (a) consistent folding and strikeslip faults are observed in the Early Tertiary red bed sequences that record the time interval between the Early Paleocene and the Late Eocene, this deformation decreasing gradually to the top of the sequence; (b) there is a remarkable absence of major normal faults affecting these sequences; and (c) north-south-trending, vertical hinges of Laramide folds are refolded as a consequence of the strike-slip of lower crust rocks. In this context we hypothesize that this heterogeneous Early Tertiary deformation can be ascribed to a general left-lateral strike-slip regime at the early stages of the Chortis block transfer to the Caribbean plate (Fig. 6). As mentioned above, this deformation regime was triggered by changes in convergence direction and subduction angles between the Farallon and North America plates. Crustal strain was distributed in a wide area along the developing transform plate boundary. The presence of the thicker and more rigid MOJB caused an inland propagation of deformation within a transpressional regime. This deformation decreased gradually as it was accommodated heterogeneously by rotation of structures and newly formed discrete shear zones. With the passage of the trench-trench-transform triple junction in the Late Eocene-Oligocene, the transform boundary was replaced by subduction. This represents a free boundary that triggered transtension inside the continental margin.

Analogue modelling of the Late Cretaceous and Early Tertiary deformation Model construction Experiments were performed at the Tectonic Modelling Laboratory of the CNR-IGG at the

COUPLING DURING LARAMIDE SHORTENING

125

Fig. 6. Cartoon showing the hypothetical model of the deformation phases, (a) Laramide deformation during the Late Cretaceous caused a wide fold-and-thrust belt, (b) During Early Tertiary transpression, new structures form around the MOJB, counterclockwise rotation and Laramide structures are refolded, rotation reaches c. 15°. Nomenclature as in Figure 1, plus Yu, Yucatan block. The position of the later exhumed Xolapa terrane is indicated (X). Arrows indicate approximate vector of convergence between Farallon and North America plates, after Engebretson et al. (1985), Meschede & Frisch (1998) and Bunge & Grand (2000).

University of Florence, Italy, and were built in a 'squeeze box' type apparatus (Fig. 7). The apparatus consists of a metallic table with a fixed wall on one side. On the opposite side there is a parallel wall that is allowed to move in different directions. Displacement of this moving wall, which is produced by electric motors and controlled (in terms of direction and velocity) by a central unit, allows simulation of normal and oblique extension, orthogonal to transpressive contraction and strike-slip deformation. Models, with dimensions of 40 cm length, 39cm width and 1.55cm average thickness, were built on the metallic table of the experimental apparatus, between the fixed and moving walls (Figs 7 & 8a). Experiments were designed to reproduce a two-phase convergence: a first

phase of orthogonal compression followed by a later stage of shortening with a lateral component in a NE direction. In order to obtain orthogonal convergence during the first phase, the moving wall of the experimental apparatus was modified by fixing an orthogonal short metallic wall at one of its extremities and producing an L-shaped wall that was allowed to move in a parallel direction with respect to the fixed wall. In this setting, the short metallic wall produced orthogonal compression of the models, with a direction of shortening that was parallel to the fixed metallic wall. The models were shortened by 42 mm (~ 11% bulk shortening) at a velocity of 6 mm h~ ! during this phase. During the second phase, the short metallic wall was removed and the moving wall was displaced in such a way as to

126

M.CERCAETAL

Fig. 7. Motorized analogue modelling apparatus used to perform the experiments.

create transpression at an angle of 15° with respect to the fixed wall. Models were shortened by 72 mm (~17% bulk oblique shortening) at a velocity of ISmmh" 1 during this second phase. Three representative models are discussed in detail in this work (see Table 1). Photographs were taken at regular time intervals with vertical and lateral illumination to observe the development and propagation of the structures. At the end of each experiment, models were covered by white sand to preserve the final topography and subsequently soaked in water to allow nondisturbed cuts of longitudinal sections.

Model rheological structure and analogue materials We constructed models that were designed to simulate a simplified two-layer vertical rheology: a brittle upper crust and a lower ductile crust (Fig. 8c & f). A parallelepiped-shaped built-in block representing the rigid block was constructed in one side of the models, adjacent to the moving wall (Fig. 8b). The parallelepiped was 25 cm long and 15 cm wide. The right and left external sides of the parallelepiped had angles of 45° and 35° with respect to the moving wall. The thickness of the brittle material was 7.5 mm, but increased to 11.5 mm within the rigid block.

Dry quartz sand with well-rounded and uniform size grains (~0.24 mm) was used to model the brittle behaviour of the upper crust. Quartz sand has a mean density of 1400 kg m~ 3 and insignificant cohesion (~70 Pa). Layers of coloured sand were sieved and sedimented as passive markers to highlight deformation in the longitudinal sections. In models Chords 02 and 03, the rigid block was simulated by thickness variation of the sand layer in order to create lateral strength heterogeneity. In model Chortis 04, the rigid block was simulated by using humid plastic clay to enhance the strength contrast between the block and 'normal crust' and to exaggerate its influence on the model deformation pattern. Plastic clay has a mean density of 2500 kg m~ 3 and unsealed high cohesion compared with the sand. A rough estimation of the clay cohesion using a shear vane tester yielded values ranging between c. 37 and 59 kPa. Model Chortis 03 was varied slightly from model Chortis 02 by lubricating the metallic table with Vaseline oil in order to reduce basal friction. To simulate the ductile behaviour of the lower crust, a homogeneous mixture of silicone polymer (Mastic Silicon Rebondissant No. 29 provided by CRC Industries, France) and sand (with a silicon: sand ratio of 5:5.5 by weight) has been utilized. The mixture has a red colour, a density ranging from 1450 to 1500kgm~ 3 , and a dynamic shear viscosity 77 = 3x105 Pa (determined at ~21°C using a coni-cylindrical viscometer).

COUPLING DURING LARAMIDE SHORTENING

127

Profile 3 clay-built block Fig. 8. Construction of the model: (a) 3D view; (b) map view; and (c) longitudinal section of models Chortis 02 and 03 and location of profiles. Strength profiles: (d) the modelled crust, (e) the modelled block constructed with sand, (f) longitudinal section of model Chortis 04, and (g) the modelled block constructed with clay.

This material exhibits near Newtonian behaviour at low strain rates such as those occurring in the experiments (<10~ 3 s~ 1 ). After construction, a grid of sand over the model surface served as a passive strain marker for the map views. The initial strength profiles of the models for maximum values of strain rate (e ~ 10~ 3 s~ 1 ) produced in the second phase are presented in Figure 8d, e and g. The properties of the materials used and a comparison with the natural properties are summarized in Table 2.

Scaling of models Scaling of a model is based on the kinematics, dynamics and geometrical similarities between the model and the natural prototype (Hubbert 1937; Ramberg 1981). Models were designed specifically to simulate the two phases of deformation that affected the south of Mexico. Thus, simplified geometry and boundary conditions were reproduced using all the geological, structural, geochronological, and geophysical information available for southern Mexico.

128

M.CERCACTAL. Table 1. Characteristics of the three experiments Chortis 02 Material used for the thicker block Cohesion contrast of the block with respect to the adjacent upper crust Lubrication at the base of model Nature of initial boundary between thicker block and lower crust

Chords 03

Chortis 04

Sand No

Sand No

Humid clay High

No Vertical

Vaseline oil Vertical

No Vertical

Models were properly scaled to Nature in such a way that 1 cm in the model is equivalent to 20 km in Nature and geometric similarity /* = ^model/mature = 5 x 10~7. In the same way, the normal stress ratio between the model and Nature must be scaled with the general stress reduction equation o* = crmodel/a-nature = p*g*/*, where the asterisks represent the ratio of the variable in the model and in Nature. This equation can be reduced to o* = p*/*, because tests on the models were conducted under normal gravity conditions, g* — 1. The mean value of density in the brittle upper crust is approximately 2750kgcm~ 3 , and the ratio p* = 0.51, so the density ratio between brittle and ductile crust (BC/DC) is approximately 0.95. With these values, a stress ratio of o* = 2.55 x 10~7 between model and Nature has been calculated. In ductile materials, viscous forces are related to dynamic shear viscosity and strain rate by o* = 8*77* or e = o* /rf> Assuming a reasonable value for the dynamic shear viscosity of the lower crust of 1021 to 1023 Pa s (e.g., Corti et al 2002; Willner et al 2002), and for the dynamic

shear viscosity of the silicone-sand mixture of 3 x 10 5 Pas, then 77* = 3 x 10~17 and e* = 8.5 x 109. The horizontal displacement velocity can be calculated from v* = vmodei/vnature = e*/*, giving v* = 4250. In the first phase of deformation, experiments represented ENE progressive Laramide shortening active during the interval from Turonian to Maastrichtian- Earliest Paleocene. Actual peak velocities computed for the Colorado Plateau in the Laramide orogeny of the Rocky Mountains are 1.5 mm a"1 (Bird 1998). However, in the Rocky Mountains the Laramide orogeny occurred within a continental plate, not involving large deformations or displacements and in a different period of time from 75 to 35 Ma (Bird 1998). In southern Mexico, the Laramide is a progressive deformation directed to the east that affects a wide area and crustal blocks of diverse composition (Salinas-Prieto et al. 2000; Cabral-Cano et al. 2000/7). Laramide shortening in southern Mexico was estimated at approximately 60 km on a balanced section in the eastern Guerrero terrane (Lang et al. 1996) and

Table 2. Model and natural parameters used in experiments Parameter*

Density, BC (kg cnT3) Cohesion, BC (kPa) Coefficient of friction, BC, JJL Density, DC (kg cm~ 3 ) Viscosity, DC, v (Pa s) Gravity, g (m s~ 2 ) Length, / (m) Stress, a (Pa) Strain rate, e (s"1) Time, 1 st phase, t (s) Time, 2nd phase, t (s) Velocity of displacement, 1st phase, v (m s"1) Velocity of displacement, 2nd phase, v (ms" 1 )

Chortis 02 and 03

Chortis 04 clay block

Nature

Model /nature ratio (referred to models Chortis 02 and 03)

1400 insignificant 0.7002 1450 3 x 105 9.81 0.01

2500 -37-59f

0.51

9.81 0.01

2750 6 x 107 0.6-0.85 -2900 1021-1023 9.81 20000

2.82 x 104 1.72 x 104 1.67 x 10~6

2.3 x 10"14 6.62 x 1014 8.19 x 1014 3.93 x 10"10

0.5 3 x 10~17 1 5 x 10"7 2.55 x 10"7 8.5 x 109 3.80 x 10"11 2.11 x 10~n 4.25 x 103

4.17 x 10"6

9.83 x 10"10

4.25 x 103

2 x 10"4 2.82 x 104 1.72 x 104 1.67 x 10"6 (6 mm IT1) 4.17 x 10"6 (15 mm IT1)

*BC, brittle crust (sand); DC, ductile crust (silicon-sand mixture). Estimated at 24% water content with a vane tester.

COUPLING DURING LARAMIDE SHORTENING

available age constraints indicate that it occurred in about 20 Ma. Using these values we obtain a rough estimation of velocity of 3 mm a"1, twice the velocity in the Rocky Mountains. In order to carry out the models in a convenient experimental time, we assumed a vmodel = 6 mmh" 1 . With this setup the experimental time decreased without a significant change in the resulting structural pattern. The second phase of deformation simulated the left-lateral transpressive regime affecting the area between the Early Paleocene and Late Eocene. Meschede & Frisch (1998) estimated over 1000km of displacement of the Chords block during the Palaeogene. Previous studies calculated deformation velocities between 54 and 56 mm a~ 1 (Herrmann etal 1994; Schaaf etal 1995) assuming a narrow plate boundary where strain is accommodated in localized shear zones. However, in a diffuse plate boundary, velocities can decrease considerably because strain is accommodated in a wide area (Gordon 1998). Furthermore, the rheological heterogeneities and mechanical anisotropy of the continental margin play an important role in the propagation and partitioning of deformation during orogenic events (Vauchez et al. 1998). With these values, a reasonable deformation velocity of approximately 31 mm a~ 1 was assumed and a model velocity of 15 mm hT1 in an ENE direction was computed.

129

the rigid block without affecting it significantly. In the models Chortis 03 and Chortis 04, increasing displacements along the existing structures were registered (Fig. 9h & 1) but no new structures were formed. A general characteristic of the deformation during the first phase is the periodic growth of the main thrusts with a spacing of ~3.5-4 cm. The model Chortis 03 also developed small backthrusts, likely as a consequence of decreasing the friction by lubricating the base with Vaseline oil. It is important to note that, due to the model design during the first phase, the movement of the wall adjacent to the rigid block caused a small boundary effect that was reflected by the slight distortion of the passive mark lines and eastward dragging of the south tip of the thrust faults in the three models. This boundary effect, however, did not affect the model results significantly.

Evolution of deformation in the second phase Figure 10 portrays the model evolution during the second phase of deformation (left-lateral transpression). In this case the bulk oblique shortening (b.o.s.) is calculated as the percentage of the moving wall displacement with respect to the resulting length of the model in the same direction (414 mm). In the models Chortis 02 and Chortis 03, during Model results the first 5.8% b.o.s. (24 mm), the transpressional deformation was accommodated by a major left Evolution of deformation during the first phase reverse-slip fault orthogonal to the first phase strucMap views and line drawings showing the evolu- tures and parallel to the moving wall. Minor faults tion of deformation during the first phase of and folds formed north and south of the main deformation are presented in Figure 9. At 3% thrust, with an angle of ~45° to the trend of this bulk shortening (3% b.s. = 12 mm), the initial structure and a length of ~2 cm (Fig. lOb & c). model deformation is manifested at the surface Pre-existing thrust faults formed during the first as north-south-striking thrusts, with vergence deformation phase started to rotate counterclocktoward the foreland, developed at about 3.5 to wise close to the moving wall during transpression; 4 cm in front of the moving wall (Fig. 9b, & j). a similar rotation pattern characterized the passive At 9% b.s., a second thrust was created 4 cm in markers on the model surface. Notably during the front of the first structure in the models Chortis second phase, deformation at the eastern side of 02 and Chortis 04 (Fig. 9c & k). In the model the block caused the development of a forelandChortis 03, the second thrust with the same verging thrust that progressively propagated to vergence was formed earlier at about 4.5% b.s., the NW following the block boundaries. In the case of model Chortis 04, the clay block 4 cm in front of the first structure. In this model, at 9% b.s. several discontinuous folds behaved as a rigid indenter in which the deforand a thrust formed between the two main struc- mation was controlled by the high strength contrast with respect to the adjacent thinner brittle tures, up to 3 cm in front of them (Fig. 9g). The models show differences in the defor- crust. No structures formed within the block mation at the end of the first deformation and the deformation propagated at the block phase, at 10.5% b.s. A third thrust with a boundaries with a higher velocity; at 3% b.o.s., regular interval of 4 cm was formed in front of all the structures showed in Figure lOj were the second structure in the model Chortis 02 already formed. Around the block, thrusts (Fig. 9d). This structure followed the shape of formed both at the clay-sand boundary and at

130

M.CERCAETAL

Fig. 9. Surface-view evolution of structures in the two phases presented in the form of a schematic table in Figures 9 and 10. The three experiments are presented in columns and consecutive east-directed shortening are presented in lines. The shaded area in line drawings corresponds to the modelled thicker crustal block. Symbols of structures as in Figure 4.

~2 cm in front of the rigid block. At the western side of the block, both these thrust sets showed a vergence towards the hinterland, whereas the thrusts at the eastern side were characterized by

a double vergence. At 5.8% b.o.s., the rigid block had moved approximately 1 cm to the NE, as indicated by distortion of the passive grid (Fig. lOj).

COUPLING DURING LARAMIDE SHORTENING

Fig. 10. Surf ace-view evolution of structures during the second phase of oblique shortening.

131

132

M. CERCAETAL.

Progressive transpression in models Chortis 02 and Chortis 03 was expressed by the continued movement on the main left re verse-slip fault and by further rotation of pre-existing structures (Fig. lOc & g). This rotation also affected the oblique faults and folds of the second phase of deformation and progressively reduced their angle to the main fault until they became aligned with its trace. Slight distortion of the passive markers and offset of the main fault suggest a small counterclockwise rotation of the rigid block. In model Chortis 04, the most important effect during this deformation interval was that movement of the rigid block a further ~1 cm to the NE caused left-lateral strike-slip structures to form. Distortion of the passive grid indicated a small clockwise rotation of the block that induced the development of an extensional basin at the west side of the block (Fig. 10k). At the end of the experiments (17.4% b.s., 72 mm of compression), the thrusts developed in models Chortis 02 and 03 at the eastern margin of the rigid block have propagated toward the NE and reached the northern part of the block; distortion of the passive markers indicates also right-lateral strike-slip faults along this margin (Fig. lOd & h). In model Chortis 04, the total displacement of the rigid block to the NE was ~3 cm and the clockwise rotation reached ~9°. A part of the block close to the moving wall was slightly uplifted and the basin in the west side of the block doubled its area. Counterclockwise rotation of the first phase thrusts was also evident on this side of the block at the end of the experiment.

related to the presence of the rigid block, deformation propagated for only a short distance in front of the moving wall in zones 2 and 3. Zone 2, corresponding to the northern part of the block, displays thrusts with opposing vergence. In the western side of the block, thrust faults related to the fold-and-thrust belt are characterized by vergence to the foreland; conversely, thrusts in the eastern side of the block show a vergence toward the moving wall (Fig. lie, g & 1). These latter structures are formed during the second phase of deformation. In zone 3, cross-sections display structures formed mainly during the second phase of deformation (Fig. lid, h & m). Longitudinal sections of model Chortis 04 (Fig. Ilk) show how the clay block acted as a rigid indenter causing high-angle reverse faults with vergence towards the block and crust uplift at its boundaries. The block remained undeformed as shown in all three sections (Fig. lid, h & m).

Summary of results The general structural pattern that resulted from the progressive deformation of the models is presented schematically in Figure 12. During the first phase, a fold-and-thrust belt with a dominant vergence towards the foreland formed parallel to the moving wall. Apart from model Chortis 04, the second phase deformation was mainly accommodated by the formation of a left reverse-slip fault orthogonal to the trend of the first phase faults, and by lateral translation and rotation of the rigid block. The higher cohesion of the clay block in model Chortis 04 prevented internal deformation and caused indentation in the adjacent crust. In Longitudinal sections all the experiments, a second system of thrusts Ten longitudinal cross-sections representative of nucleated at the eastern margin of the thicker the final stage of model deformation are shown in crustal block and then propagated northwestward Figure 11 a, c and i. These sections highlight the in domain I (Fig. 12). With increasing deformation, different influence of the two phases of deformation a second fault system was observed in the west side in different parts of the model, defining three dis- of the rigid block, domain II. Vertical-axis countertinct regions: (1) a zone extending northward clockwise rotation of structures was observed in all beyond the rigid block, affected only by the first three models in the interference zone between the phase of deformation (Fig. 1 Ib, f & j); (2) a zone two phases of deformation. Additionally, the clay corresponding to the northern part of the block block in model Chortis 04 rotated clockwise affected by structures belonging to both the two during progressive deformation; this rotation deterphases of deformation (Fig. lie, g & 1); and (3) a mined the development of an extensional basin and southern zone close to the contact between the NE-striking left-lateral faults in domain II. rigid block and the second phase moving wall characterized by structures formed mainly during the second phase of deformation (Fig. 1 Id, h & m). Qualitative comparison of model results In zone 1, shortening resulted in the develop- with the geology ment of thrusts and box-type folds; thrusts with a prevailing vergence to the foreland, propagated Limitations of modelling up to ~9 cm in front of the moving wall (Fig. 1 Ib, Many natural parameters concerning the rhef & j). Because of the high strength contrast ology and the boundary conditions of the

COUPLING DURING LARAMIDE SHORTENING

Fig. 11. Photographs of longitudinal sections.

133

134

M. CERCAETAL.

Fig. 12. Simplified map of resulting domains of deformation. Arrows indicate the direction of convergence.

deformation process under investigation are not easy to obtain and frequently data in the literature are scarce, as is the case in southern Mexico. As a consequence, analogue modelling necessarily simplifies the geometry and the rheology of the complex natural process and these simplifications have to be made explicit before comparing the results with the geology. Geometrical simplifications of the current models involve the simulation of the Laramide orogeny that caused deformation in a wide fold-and-thrust belt in southern Mexico. This deformation phase has been attributed to mechanical coupling between a subhorizontally subducted slab and an overriding continental crust (Dickinson et al 1988; Bird 1998). However, it has been observed that analogue models shortened by advancing a rigid vertical boundary laterally simulate most of the characteristics of fold-and-thrust belts (e.g., Bonini 2001). In the second phase, the deformation is attributed to the motion of the Chords block after its partial detachment from North America in the Early Tertiary (Herrmann et al 1994). The shear zone bounding the Xolapa complex to the north has an ESE strike, but, as structures observed in the field are mostly compatible with a transpressive regime, the movement of the wall was made to simulate an ENE direction of the contraction, and variations in the plate boundary through time were not considered. During the experiments, the compressive stresses causing model deformation are transmitted from the rigid moving wall, thus simulating a lateral transmission of forces from the plate boundaries. This

represents a simplification of the natural process, where forces causing deformation of the lithosphere at plate boundaries may be transmitted vertically from below, driven by deformation of the mantle (Teyssier & Tikoff 1998). Additionally, in all the models, the rigid block was attached to the moving wall during transpression, preventing important block rotations likely to have occurred in the natural prototype. From a rheological point of view, the simple two-layer structure of the model, which is intended to simulate the crust only, is a further simplification of the natural situation. The experimental series was designed to investigate the deformation around a strong (rigid) crustal block (MOJB) embedded within a 'normal' continental crust. Since the crustal strength resides in the upper brittle layer, our model set-up considered a uniformly thick continental crust characterized by a region with thicker brittle crust. This implies that in the models the ductile crust below the rigid block is thinner than elsewhere. Unless the crust and mantle are decoupled, the compression of two adjacent crustal sections with different thicknesses may result in the indentation of the upper mantle of the thinner block into the less rigid lower crust of the thicker block (Harry et al 1995). In addition, the cohesion of clay used in model Chords 04 is not properly scaled and clearly exceeds the rigidity of the natural prototype, so that the clay block behaved as an indenter, emphasizing deformations at block boundaries. For this reason the results can only be qualitavely compared to the structures in the field. Several factors can have an influence on the structures formed by deformation, such as erosion and deposition in basins, pre-existing structures, effect of pore pressure in the growth and propagation of structures, thermal evolution or isostasy effects, none of which were considered in the modelling. Nevertheless, despite the abovementioned simplifications, comparisons of model results with Nature were useful in understanding the structural evolution. Comparison with Nature Although a single model cannot explain all the structural complexity observed in Nature, the combined results of our experiments simulated most of the styles of deformation and largescale structures observed around the rigid crustal block of the MOJB. Thus, these results suggest that the processes affecting both model and natural prototype were similar. We emphasize that our models are pertinent in a time interval from ~88 to ~36Ma. Line drawings of

COUPLING DURING LARAMIDE SHORTENING

models compared with a schematic structural map of the south of Mexico are presented in Figure 13. The fold-and-thrust belt (Fig. 13, structure 1) of southern Mexico has been associated with the amalgamation of tectonic blocks during the Mesozoic or the Early Tertiary, the time of the Laramide episode. Recent microstructural data in the eastern Guerrero terrane suggest that the

135

Laramide structures have a double vergence and they have been interpreted in terms of a progressive ductile shear (Salinas-Prieto et al. 2000). During our experiments, double vergence was modelled and was accentuated by the presence of the rigid block. The first effect of the second phase of deformation is counterclockwise rotation of the thrusts and folds in the southwestern part of

Fig. 13. Comparison of models with the natural prototype. The natural prototype structures are based mainly on Campa & Coney (1983), Sedlock et al. (1993), Ortega-Gutierrez et al (1999), Elias-Herrera & Ortega-Gutierrez (2002), Ham-Wong (1981) and our own field data.

136

M. CERCA£TAL.

the model (Fig. 13, structure 2). Similar effects have been documented in the southern part of the Guerrero-Morelos platform (Cerca & Ferrari 2001). Other evidence of Tertiary counterclockwise vertical-axis rotations in this area has been inferred from the palaeomagnetic data from Cretaceous carbonate sequences (Molina-Garza et al 2003). Vertical-axis rotations and lateral translation of the high-strength block in the model mirror complex structures and thrusting of the rigid block over the adjacent crust at Papalutla (Fig. 13, structure 3), consistent with the geometry of the Papalutla thrust and related deformation in the Early Tertiary basins to the NW (Cerca & Ferrari 2001). Progressive strain at the eastern margin of the rigid block produced a thrust that propagated to the NW closely following the geometry of the block. This structure strikingly resembles the geometry and the kinematics of the Vista Hermosa fault (Fig. 13, structure 4), which has thrust Palaeozoic schists over Jurassic rocks (Sedlock et al 1993). To the north of the rigid block in the experiments, uplift and arcuate folding of the adjacent part of the model were observed. This pattern of deformation, clearly influenced by the geometry of the rigid block, was emphasized in the model with the clay block. Similar structures were observed in the natural prototype in the Tentzo Ranges (Fig. 13, structure 5) where folds of Cretaceous carbonates define an arc convex toward the north. Although most of these folds were produced by the decollement of the carbonate succession, this process was likely triggered by the uplift of the basement as simulated in the model. As in the case of the Papalutla fault, the Tentzo Ranges were considered to be produced during the Laramide orogeny (Monroy & Sosa 1984). However, their anomalous orientation with respect to the general trend of the Laramide structures has not hitherto been explained. One obvious difference between analogue models and southern Mexico is that the structures observed within the MOJB did not develop in the models. These structures can be explained as reactivation of discontinuities existing before the Laramide event. One major example is the Oaxaca fault, which exhibits a complex history of lateral, inverse and normal movements beginning at least from the Jurassic or even the Palaeozoic time (Alaniz-Alvarez et al. 1994, 1996). Only tight folds and strike-slip faults affecting Tertiary red beds (Fig. 13, structure 6) resemble structures formed in the models. Finally, in the case of the model Chortis 04, left-lateral strikeslip and a basin are formed in the southwestern side of the block (Fig. 13, structure 7) coincident

with the deposition of Tertiary red beds in the eastern Guerrero state. Vertical coupling and decoupling in the crust of southern Mexico during the Late Cretaceous and the Tertiary Our new fieldwork linked with previous research has recognized three deformation phases in southern Mexico: Late Cretaceous (Laramide) shortening, Early Tertiary left-lateral transpression, and post-Eocene transtension. A series of analogue models have been used to simulate the role of a rigid and thicker block (MOJB) within the upper crust during the first two deformation phases. The design of the models implies that the resulting structural pattern is mainly controlled by the strength of the upper brittle crust and the forces applied at the vertical boundaries of the MOJB. Although these conditions are given a priori, we believe that this could be the real case. Indeed it is likely that during the Laramide orogeny the mantle lithosphere was removed or at least weakened by the subhorizontal subduction of the Farallon plate. If this is the case, in the following stage the upper crust remained the most rigid part of the whole system. This view supports the claim of Jackson (2002) that the detailed patterns of surface faulting in erogenic zones are predominantly controlled by the strength of the upper crustal blocks and by the faults bounding them rather than by forces applied at the base of the lithosphere. In addition, in the models, decoupling between ductile and brittle layers was enhanced by the presence of the block. The higher strength and thickness of the block increase the mechanical contrast between brittle and ductile layers and hence the decoupling between lower and upper layers. This causes a propagation of strain in a wide area around the block during the second deformation phase. Thus, in our models, coupling or decoupling between the ductile and brittle layers are determined by the thickness and the rigidity of the upper crustal layer. In Nature, southern Mexico was subject to a general left-lateral strike slip regime along the developing Caribbean-North America plate boundary. During this Early Tertiary phase, strain in the lower crust should have been distributed homogeneously in the deformation zone between the two plates. By contrast, the deformation observed in the upper crust was focused on a small number of structures around the MOJB. This radically different pattern of deformation implies a small degree of coupling between upper and lower crust.

COUPLING DURING LARAMIDE SHORTENING

Conclusions We have performed a series of analogue modelling experiments to simulate the effect of two deformation phases on a layered brittle-ductile modelled crust incorporating a thicker upper crustal block. The block was constructed with either sand or clay to compare the mechanical anisotropy effects of a high contrast in strength and cohesion. Models reproduced most of the structures observed around the MOJB in Nature, suggesting a close similarity in deformation processes between Nature and model, and permit the reinterpretation of several key features of the Early Tertiary geological evolution of southern Mexico. We identify structures consistent with a transpressive regime during the second phase, which interfere with the structures formed previously in the Laramide event. We show that lateral partitioning of phase 2 strain is important north of the Xolapa complex, in contrast with previous interpretations that assign these structures to the Laramide deformation. The models predict a Tertiary motion along the Vista Hermosa fault, and that the structure propagated towards the NW following the margin of the MOJB. In the models, the propagation of deformation to the north is related to mechanical contrasts in strength and cohesion of the block with respect to the adjacent modelled crust. Structures formed in the tectonic interference zone present counterclockwise vertical-axis rotations consistent with the field evidence and palaeomagnetic data. Structures within the MOJB observed in Nature do not develop in the model. This is likely due to reactivation of pre-existing structures that influenced the deformation within the block, a factor not considered in the model. Without pre-existing structures the strength of the block increases as a function of the thickness. The experiments show that most of the structures observed in Nature can be reproduced using relatively few parameters and a simple two-layer analogue model. In our models, the degree of coupling between the ductile and brittle layers is controlled by the thickness and the rigidity of the upper crustal layer. The Early Tertiary structural patterns observed in southern Mexico around the MOJB suggest that it behaved similarly to the model. This research was supported by grant CONACyT 32509-T (to LF) and CONACYT-CNR bilateral grant. We thank Giovanna Moratti, Chiara Del Ventisette and Domenico Montanari for their support and help during the laboratory work. Guido Schreurs, John Grocott, Dora CarreonFreyre, Susana Alaniz and an anonymous reviewer provided constructive criticisms that enormously improved

137

this work. MC also thanks CONACYT, whose grants permitted him to pursue his PhD, and a research stay in Italy.

References ALANIZ-ALVAREZ, S.A., NIETO-SAMANIEGO, A.F. & ORTEGA-GUTIERREZ, F. 1994. Structural evolution of the Sierra de Juarez mylonitic complex, State of Oaxaca, Mexico. Revista Mexicana de Ciencias Geological, II, 147-156. ALANIZ-ALVAREZ, S.A., VAN DER HEYDEN, P., NIETOSAMANIEGO, A.F. & ORTEGA-GUTIERREZ, F. 1996. Radiometric and kinematic evidence for Middle Jurassic strike-slip faulting in southern Mexico related to the opening of the Gulf of Mexico. Geology, 24 (5), 443-446. ALANIZ-ALVAREZ, S.A., NIETO-SAMANIEGO, A.F., MORAN-ZENTENO, D.J. & ALVA-ALDAVE, L. 2002. Rhyolitic volcanism in extension zone associated with strike-slip tectonics in the Taxco region, southern Mexico. Journal of Volcanology and Geothermal Research, 118, 1-14. BIRD, P. 1998. Kinematic history of the Laramide orogeny in latitudes 35°—49° N, western United States. Tectonics, 17 (5), 780-801. BONINI, M. 2001. Passive roof thrusting and forelandward fold propagation in scaled brittle-ductile physical models of thrust wedges. Journal of Geophysical Research, 48 (B2), 2291-2311. BUNGE, H.P. & GRAND, S.T. 2000. Mesozoic platemotion history below the northeast Pacific Ocean from seismic images of the subducted Farallon slab. Nature, 405, 337-340. CABRAL-CANO, E., LANG, H.R. & HARRISON, C.G.A. 2000a. Stratigraphic assessment of the ArceliaTeloloapan area, southern Mexico: implications for southern Mexico's post-Neocomian tectonic evolution. Journal of South American Earth Sciences, 13, 443-457. CABRAL-CANO, E., DRAPER, G., LANG, H.R. & HARRISON, C.G.A. 2000/?. Constraining the Late Mesozoic and Early Tertiary tectonic evolution of southern Mexico: structure and deformation history of the Tierra Caliente region, southern Mexico. Journal of Geology, 108, 427-446. CAMPA, M.F. 1978. La evolucion tectonica de Tierra Caliente, Guerrero. Boletin de la Sociedad Geologica Mexicana, 39, 52-64. CAMPA, U.M.F. & CONEY, P.J. 1983. Tectonostratigraphic terranes and mineral resource distributions in Mexico. Canadian Journal of Earth Sciences, 20, 1040-1051. CAMPOS-ENRIQUEZ, J.O. & SANCHEZ-ZAMORA, O. 2000. Crustal structure across southern Mexico inferred from gravity data. Journal of South American Earth Sciences, 13, 479—489. CERCA, M. & FERRARI, L. 2001. Vertically decoupled Paleocene-Eocene wrenching in southern Mexico and its possible relation with the activation of the Caribbean-North America boundary. Eos Trans. AGU, 82 (47), Fall Meet. Suppl., Abstract T12D-0938.

138

M. CERCAETAL

CERCA, M., FERRARI, L. & LOPEZ-MARTINEZ, M. 2003. Early Tertiary deformation in southern Mexico and its relation to the transfer of the Chords block from the North America to the Caribbean plate. Geological Society of America Cordilleran Section Meeting 2003, Abstract T8 33-9. CONSEJO DE RECURSOS MINERALES 1998. Carta geologico-minera y geoquimica de la Hoja San Miguel Talea de Castro, Escala 1:50,000 (E14 D-39), Consejo de Recursos Minerales, Pachuca, Hgo., Mexico. CONSEJO DE RECURSOS MINERALES 2001. Carta geologico-minera y geoquimica de la Hoja Oaxaca, Escala 1:250,000 (E14-9), Consejo de Recursos Minerales, Pachuca, Hgo., Mexico. CORTI, G., BONINI, M., MAZZARINI, F., BOCCALETTI, M., INNOCENTI, F., MANETTI, P., MULUGETA, G. & SOKOUTIS, D. 2002. Magma-induced strain localization in centrifuge models of transfer zones. Tectonophysics, 348, 205-218. DE CSERNA, Z., ORTEGA-GUTIERREZ, F. & PALACIOSNIETO, M. 1980. Reconocimiento geologico de la parte central de la cuenca del alto Rio Balsas, estados de Guerrero y Puebla. In: Sociedad Geologica Mexicana, Libro Guia de la Excursion Geologica a la Parte Central de la Cuenca del Alto Rio Balsas, V Convention Geologica Nacional, 2-33. DICKINSON, W.R., KLUTE, M.A., HAYES, M.J., JANECKE, S.U., LUNDIN, E.R., MCKITTRICK, M.A. & OLIVARES, M.D. 1988. Paleogeographic and paleotectonic setting of Laramide sedimentary basins in the central Rocky Mountain region. Geological Society of America Bulletin, 100, 1023-1039. DUCEA, M., SHOEMAKER, S., GEHRELS, G., VERVOORT, J. & Ruiz, J. 2003. Zircon U-Pb geochronology constraints on the magmatic and tectonic evolution of the Xolapa complex, southern Mexico. Geological Society of America Cordilleran Section Meeting 2003, Abstract with Programs. ELIAS-HERRERA, M. & ORTEGA-GUTIERREZ, F. 2002. Caltepec fault zone: an Early Permian dextral transpressional boundary between the Proterozoic Oaxacan and Paleozoic Acatlan Complexes, southern Mexico, and regional tectonic implications. Tectonics, 21 (3), 1-18. ENGEBRETSON, D.C., Cox, A. & GORDON, R.G. 1985. Relative plate motions between oceanic and continental plates in the Pacific basin. Geological Society of America Special Paper, 206, 1-64. FRIES, C., JR. 1960. Geologia del Estado de Morelos y de Partes Adyacentes de Mexico y Guerrero, Region Central Meridional de Mexico. Boletin de la Sociedad Geologica Mexicana, 60. GARCIA-PEREZ, F. & URRUTIA-FUCUGAUCHI, J. 1997. Crustal structure of the Arteaga Complex, Michoacan, southern Mexico, from gravity and magnetics. Geoisica Internacional, 36 (4), 235-244. GORDON, R.G. 1998. The plate tectonic approximation: plate nonrigidity, diffuse plate boundaries, and global plate reconstructions. Annual Review of Earth and Planetary Sciences, 26, 615-642. HAM-WONG, J.M. 1981. Prospecto: Guelatao. Informe Geologico No. 799, Zona Sur. Petroleos Mexicanos.

HARRY, D.L., OLDOW, J.S. & SAWYER, D.S. 1995. The growth of orogenic belts and the role of crustal heterogeneities in decollement tectonics. Geological Society of America Bulletin, 107, 14111426. HERNANDEZ-ROMANO, U., AGUILERA-FRANCO, N., MARTINEZ-MEDRANO, M. & BARCELO-DUARTE, J. 1997. Guerrero-Morelos Platform drowning at the Cenomanian-Turonian boundary, Huitziltepec area, Guerrero State, southern Mexico. Cretaceous Research, 18,661-686. HERRMANN, U.R., NELSON, B.K. & RATSCHBACHER, L. 1994. The origin of a terrane: U/Pb zircon geochronology and tectonic evolution of the Xolapa Complex (southern Mexico). Tectonics, 13, 455-474. HUBBERT, M.K. 1937. Theory of scale models as applied to the study of geologic structures. Bulletin of the Geological Society of America, 48, 1459-1520. JACKSON, J. 2002. Strength of the continental lithosphere: time to abandon the jelly sandwich? GSA Today, September, 4-9. JORDING, A., FERRARI, L., ARZATE, J. & JODICKE, H. 2000. Crustal variations and terranes boundaries in southern Mexico as imaged by magnetotelluric transfer functions. Tectonophysics, 327, 1-13. LANG, H.R. & FRERICHS, W.E. 1998. New planktic foraminiferal data documenting Conician age for Laramide Orogeny onset and pelooceanography in southern Mexico. Journal of Geology, 106, 635-640. LANG, H.R., BARROS, J.A., CABRAL-CANO, E., DRAPER, G., HARRISON, C.G.A., JANSMA, P.E. & JOHNSON, C.A. 1996. Terrane deletion in northern Guerrero state. Geofisica Internacional, 35 (4), 349-359. MESCHEDE, M. & FRISCH, W. 1998. A plate-tectonic model for the Mesozoic and Early Cenozoic history of the Caribbean plate. Tectonophysics, 296, 269-291. MESCHEDE, M., FRISCH, W., HERRMANN, U.R. & RATSCHBACHER, L. 1996. Stress transmission across an active plate boundary: an example from southern Mexico. Tectonophysics, 266, 81-100. MEZA-FIGUEROA, D., VALENCIA-GOMEZ, V., OCHOA-LANDIN, L., PEREZ-SEGURA, E. & DIAZ-SALGADO, C. 2001. 40Ar/39Ar geochronology and geochemistry of Laramidic plutons from the Mezcala mining district, Guerrero, Mexico. Actas INAGEQ, 1 (1), 21. MILLS, R.A. 1998. Carbonate detritus and mylonite zones in Guerrero, Mexico and northern Honduras: new evidence for detachment of the Chortis block from southern Mexico. Journal of South American Earth Sciences, 11 (3), 291-307. MOLINA-GARZA, R.S., BOHNEL, H. & HERNANDEZ, T.

2003. Paleomagnetism of the Cretaceous Morelos and Mezcala formations, southern Mexico. Tectonophysics, 361, 301-317. MONROY, M. & SOSA, A. 1984. Geologia de la Sierra del Tentzo, Puebla, borde norte del terreno

COUPLING DURING LARAMIDE SHORTENING Mixteco. Sociedad Geologica Mexicana, Boletin, 45 (1 -2), 43-72. MORAN-ZENTENO, DJ. 1992. Investigaciones isotopicas de Rb-Sr y Sm-Nd en rocas cristalinas de la region Tierra Colorada-Acapulco-Cruz Grande, Estado de Guerrero. PhD thesis, UNAM. MORAN-ZENTENO, DJ., CORONA-CHAVEZ, P. & TOLSON, G. 1996. Uplift and subduction erosion in southwestern Mexico since the Oligocene: pluton geobarometry constraints. Earth and Planetary Science Letters, 141, 51-65. MORAN-ZENTENO, DJ., TOLSON, G., et al 1999. Tertiary arc-magmatism of the Sierra Madre del Sur, Mexico, and its transition to the volcanic activity of the Trans-Mexican Volcanic Belt. Journal of South American Earth Sciences, 12, 513-535. MURILLO, G., TORRES, R. & NAVARRETE, O. 1992. El macizo de la Mixtequita: redefinicion. XI Convencion Geologica NacionaL Sociedad Geologica Mexicana, Libro de Resumenes, 131. ORTEGA-GUTIERREZ, F. 1980. Rocas volcanicas del Maestrichtiano en el area de San Juan Tetelcingo, Estado de Guerrero. In: Sociedad Geologica Mexicana, Libro Guia de la Excursion Geologica a la Parte Central de la Cuenca del Alto Rio Balsas, V Convencion Geologica Nacional, 34-38. ORTEGA-GUTIERREZ, F., MITRE-SALAZAR, L.M., ROLDAN-QUINTANA, J., SANCHEZ-RUBIO, G. &

DE LA FUENTE, M. 1990. North America Continent—Ocean Transects Program, Transect H-3: Acapulco Trench to the Gulf of Mexico Across Southern Mexico. Geological Society of America. ORTEGA-GUTIERREZ, F., ELIAS-HERRERA, M., REYESSALAS, M., MACIAS-ROMO, C. & LOPEZ, R. 1999. Late Ordovician-Early Silurian continental collisional orogeny in southern Mexico and its bearing on Gondwana-Laurentia connections. Geology, 27, 719-722. ORTEGA-GUTIERREZ, F.L., Mitre-Salazar, L.M., ROLDAN-QUINTANA,

J.;

ARANDA-GOMEZ,

J.J.,

MORAN-ZENTENO, D., ALANIZ-ALVAREZ, S.A. & NIETO-SAMANIEGO, A.F. 1992. Carta Geologica de la Republica Mexicana, 5a edicion 1: 2000000 con texto explicativo. Instituto de Geologia, Universidad Nacional Autonoma de Mexico, Consejo de Recursos Minerales. RAMBERG, H. 1981. Gravity, Deformation and Earth's Crust. Academic Press, San Diego, CA, 452. RATSCHBACHER, L., RILLER, U., MESCHEDE, M., HERRMANN, U. & FRISCH, W. 1991. Second look at suspect terranes in southern Mexico. Geology, 19, 1233-1236.

139

RILLER, U., RATSCHBACHER, L. & FRISCH, W. 1992. Left-lateral transtension along the Tierra Colorada deformation zone, northern margin of the Xolapa magmatic arc, of southern Mexico. Journal of South American Earth Sciences, 5, 237-249. SALINAS-PRIETO, J.C., MONOD, O. & FAURE, M. 2000. Ductile deformations of opposite vergence in the eastern part of the Guerrero Terrane (SW Mexico). Journal of South American Earth Sciences, 13, 389-402. SCHAAF, P., MORAN-ZENTENO, D., HERNANDEZBERNAL, M.S., SOLIS-PICHARDO, G., TOLSON, G. & KOHLER, H. 1995. Paleogene continental margin truncation in southwestern Mexico: geochronological evidence. Tectonics, 14,1339-1350. SEDLOCK, R.L., ORTEGA-GUTIERREZ, F. & SPEED, R.C. 1993. Tectonostratigraphic terranes and tectonic evolution of Mexico. Geological Society of America Special Paper, 278, 153. TEYSSIER, C. & TIKOFF, B. 1998. Strike-slip partitioned transpression of the San Andreas fault system: a lithospheric scale approach. In: Continental Transpression and Transtension Tectonics. HOLDSWORTH, R.E., STRACHAN, R.A. & DEWEY,

J.F. (eds) Geological Society, London, Special Publications, 135, 143-158. URRUTIA-FUCUGAUCHI, J. & FLORES-RUIZ, J.H. 1996. Bouguer anomalies and regional crustal structure in central Mexico. International Geology Review, 38, 176-194. URRUTIA-FUCUGAUCHI, J. & FERRUSQUIA-VILLAFRANCA, I. 2001. Paleomagnetic results for the Middle-Miocene continental Suchilquitongo Formation, Valley of Oaxaca, southeastern Mexico. Geofisica Internacional, 40, 191-206. VALDEZ, C., MOONEY, W., et al. 1986. Crustal structure of Oaxaca, Mexico, from seismic refraction measurements. Bulletin of the Seismological Society of America, 76, 547-563. VAUCHEZ, A., TOMMASI, A. & BARROUL, G. 1998. Rheological heterogeneity, mechanical anisotropy and deformation of the continental lithosphere. Tectonophysics, 296, 61-86. WEBER, B. & KOHLER, H. 1999. Sm-Nd, Rb-Sr and U—Pb isotope geochronology of a Grenville terrane in southern Mexico: origin and geologic history of the Guichicovi complex. Precambrian Research, 96, 245-262. WILLNER, A.P., SEBAZUNGU, E., GERYA, T.V., MARESCH, W.V. & KROHE, A. 2002. Numerical modeling of PT-paths related to rapid exhumation of high pressure rocks from the crustal root in the Variscan Erzgebirge Dome (Saxony/Germany). Journal of Geodynamics, 33, 281—314.

This page intentionally left blank

Distributed strike-slip faulting, block rotation and possible intracrustal vertical decoupling in the convergent zone of SW Japan OLIVIER FABBRI1, KAZUMASA IWAMURA2, SATOSHI MATSUNAGA3, GUILHEM COROMINA1 & YUJI KANAORI4 1 EA 2642, Universite de Franche-Comte, 16 route de Gray, 25030 Besangon Cedex, France (e-mail: [email protected]) 2 Iwakuni High School, Iwakuni 741-0082, Japan 3 Nissaku Co., Joetsu 950-2181, Japan 4 Department of Geology, Faculty of Science, Yamaguchi University, Yamaguchi 753-8512, Japan Abstract: Between the Median Tectonic Line (MTL) and the Japan Sea, the western Chugoku region of SW Japan is cut by a series of N45°E first-order faults and oblique (N60°-N170°E) second-order faults. This fault network, probably formed during Late Cretaceous-Palaeocene times (70-60 Ma), defines a regional block structure. PrePlio-Quaternary kinematical indicators suggest left-lateral motion along the first-order faults and right-lateral motion along some of the second-order faults. Geomorphological evidence and earthquake focal mechanisms indicate that Plio-Quaternary slip senses are opposite to Pre-Plio-Quaternary ones. The overall fault pattern is geometrically and kinematically similar to patterns obtained by experimental modelling of simple shear deformation distributed at the base of a brittle layer analogue over its entire width. This similarity suggests the possibility of a midcrustal, flat-lying partial attachment zone which could have controlled the formation of the western Chugoku fault network in Cretaceous to Palaeocene times. The zone, presently inactive, could correspond to the 'proto-MTL', a low-angle fault recently imaged by seismic reflection studies and whose trace approximately coincides with the present-day MTL. Reactivation of the system occurred twice after its formation: firstly in Miocene times, during the opening of the Japan Sea and concomitant clockwise rotation of the entire SW Japan arc; and secondly in Late Pliocene to Quaternary times, after a shift of the relative direction of convergence between the Philippine and Eurasia plates. Unlike the first reactivation, the second reactivation led to an inversion of the sense of slip along the faults.

Wrench tectonics can be localized along a single strike-slip fault but can also be diffuse and encompass large areas whose widths can exceed several tens or hundreds of kilometres (e.g. Moore 1979). Diffuse wrench zones are composed of crustal blocks and slivers delimited by vertical faults. The mechanism of internal deformation of diffuse wrench zones typically involves vertical-axis rotations of blocks and concomitant strike-slip along the block-bounding faults, the so-called 'bookshelf mechanism of Mandl (1987). Examples of analyses of internal deformation of diffuse wrench zones are numerous and often rely on palaeomagnetic investigations (Luyendyk et al. 1980; McKenzie & Jackson 1983; Ron et al. 1984; Nicholson et al. 1986; Martel et al. 1988; Kanaori 1990;

Kanaori et al. 1990, 1992; Brown et al. 1993; Dickinson 1996; Taylor et al. 1998; Kuhn & Reuther 1999; Randall et al. 2001; Grocott & Taylor 2002). Experimental analogue modelling also provide insights into the geometry and kinematics of such diffuse wrench zones (Hempton & Neher 1986; Schreurs 1994). A key question regarding distributed strikeslip faulting and associated crustal blocks and slivers concerns the downward extension of these structures. Do they progressively disappear when reaching the ductile lower crust, or are they abruptly separated from the underlying crust (whatever its rheology may be) by a horizontal or gently dipping detachment or partial attachment zone? A detachment zone would allow a complete vertical decoupling, whereas a partial

From: GROCOTT, J., MCCAFFREY, K. J. W., TAYLOR, G. & TIKOFF, B. (eds) 2004. Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 141-165. 0305-8719/047$ 15 © The Geological Society of London 2004.

142

O.FABBR1ETAL.

attachment zone would still allow some coupling and transmission of the stresses from below the zone to above it (Tikoff et al 2002). Midcrustal flat-lying detachment or partial attachment zones could correspond to deep reflectors imaged on seismic reflection profiles obtained across some large strike-slip faults (Lemiszki & Brown 1988; Brocher et al. 1994). It is interesting to note that mechanical models based on these geometries appear compatible with measured strains (Biirgmann 1997). The aim of this chapter is to provide an example of a distributed strike-slip fault system located in the convergent zone of the SW Japan arc. Geometrical and kinematical similarities with fault patterns obtained by experimental analogue modelling suggest a relationship between the formation of the upper crust fault network and the activity of an ancient, presently inactive, mid-crustal low-angle partial attachment zone.

General geology of the SW Japan arc The Itoigawa-Shizuoka Tectonic Line (ISTL on Fig. 1) is classically regarded as the boundary between the SW and NE Japan arcs. The SW Japan arc is further divided by the Median Tectonic Line (MTL) into an Outer Zone on the Pacific Ocean side, and an Inner Zone on the Japan Sea side. Both Inner and Outer Zones consist of complex superpositions or juxtapositions of arc-parallel elongated tectonic belts ranging in age from Palaeozoic to OligoMiocene, with a clear southeastward younging polarity from the Japan Sea side to the Pacific Ocean side (Taira et al. 1982; Ichikawa et al. 1990; Isozaki et al 1990; Isozaki 1996; Nakajima 1997). The Outer Zone is well exposed in the Shikoku Island where it is relatively devoid of plutonic intrusion or volcanic cover (Fig. 2). It consists of a series of elongated belts, which are, from the Pacific coast to the MTL: the Shimanto belt, a Cretaceous-Palaeogene accretionary complex (Taira et al. 1980, 1982; Hibbard & Karig 1987, 1990; Mackenzie et al 1987; Taira & Ogawa 1988; Agar et al 1989; Awan & Kimura 1996); the southern Chichibu (also called Sanbosan) belt, a Jurassic accretionary complex (Matsuoka & Yao 1990; Matsuoka 1992); the Kurosegawa belt, a Permian accretionary complex (Murata 1982; Maruyama et al 1984; Yoshikura et al 1990; Isozaki & Itaya 1991); the northern Chichibu belt, a Late Triassic to Jurassic accretionary complex (Hada & Kurimoto 1990); and the Cretaceous high-P/T metamorphic Sanbagawa belt (Hara et al 1990,

1992; Wallis & Banno 1990; Takasu et al 1994; Dallmeyer et al 1995). The Inner Zone shows a more complicated geological structure than the Outer Zone, partly because of igneous intrusions and volcanic deposits which make mapping difficult. The geological basement of the Inner Zone consists of a Carboniferous-Permian to Jurassic infrastructure extensively covered by Cretaceous volcanic or volcano-sedimentary deposits (Fig. 1). The infrastructure and the volcanic cover are intruded or covered by Late Cretaceous-Palaeogene igneous complexes and by isolated Neogene to Quaternary volcanic or sedimentary deposits. The infrastructure consists of two superposed accretionary complexes. The older one is composed of a stack of nappes involving Carboniferous to Lower Triassic metamorphic and sedimentary rocks along with mafic or ultramafic tectonic lenses (Faure & Charvet 1987; Nishimura 1990). The metamorphic rocks, which are exposed in the Sangun-Renge and Suo belts, are of high-P/T type and their radiometric ages (mainly K-Ar or Rb-Sr white mica ages from pelitic or psammitic schists, and some K-Ar actinolite and barroisite ages from basic schists) range between 330 and 160 Ma (Nishimura 1990, 1998). The stack of nappes is locally unconformably overlain by weakly deformed Upper Triassic to Jurassic shallow-water deposits. This complex technically overlies a Jurassic accretionary complex involving non- or weakly metamorphosed sedimentary rocks including exotic blocks whose ages range from Permian to Jurassic (Caridroit etal 1986; Mizutani 1990). After the tectonic superposition, large parts of the southernmost and structurally lowest part of the Inner Zone, that is the Jurassic accretionary complex, were affected by the low-P/7 Ryoke metamorphism (Fig. 2). This regional metamorphism comprises low-pressure facies series of andalusite-sillimanite type with a migmatite zone in the high-grade part of the 30-50 km wide belt. The Ryoke metamorphism occurs superimposed on the Jurassic accretionary complex, except in Kyushu, where the SangunRenge belt is also affected. The thermal axis of the plutono-metamorphic Ryoke belt is obliquely cut by the MTL. Details on the Ryoke metamorphism are given by Nakajima et al (1992) and Brown (1998). In Shikoku (Fig. 2), but also westwards (in Kyushu) and eastwards (Kii peninsula), the southern rim of the Ryoke belt is unconformably covered by Mid- to Late Cretaceous marine clastic deposits forming elongated basins parallel to, and cut by, the MTL (Miyata 1990). In Shikoku, these deposits constitute the Izumi

DISTRIBUTED STRIKE-SLIP FAULTING IN JAPAN

143

Fig. 1. Tectonic framework of the study area. Plate boundaries are indicated by dashed lines with black triangles (after Research Group for Active Faults of Japan 1991). Tectonic plates are the Eurasia plate, Philippine Sea plate, Pacific plate and North America (NAM) plate. Major faults are the Sikhote-Alin fault system, the Yangsan fault system (YFS), the Tsushima fault system (TFS), the Median Tectonic Line (MTL), the Southern Japan Sea fault zone (SJSFZ), the Itoigawa-Shizuoka Tectonic Line (ISTL) and the Tanakura Tectonic Line (TTL). The trace of the Sikhote-Alin fault system is after Geological Survey of Japan (I992a). The trace of the SJSFZ is after Itoh et al (2002). QVF is the Quaternary Volcanic Front for SW Japan (after Research Group for Active Faults of Japan 1991).

Fig. 2. Geology of the Shikoku-western Honshu region (simplified after Geological Survey of Japan 1992b). Arrows on the Japan Sea side indicate palaeodeclinations deduced from palaeomagnetic measurements carried out on Cretaceous to Miocene rocks (Otofuji & Matsuda 1983, 1984, 1987; Otofuji et al. 1985, 1991). Rocks younger than c. 14 Ma do not present deflected declinations. The inset NW of Hiroshima is for Figure 3. The dotted line along the Seto Inland Sea is the northern limit of the Ryoke low-P/7metamorphic belt.

DISTRIBUTED STRIKE-SLIP FAULTING IN JAPAN

basin and are Campanian-Maastrichtian in age. Their thickness is exceptionally high, suggesting that they were deposited in wrench-faulted basins. As a whole, there is an eastward younging of the age of the strata, interpreted as reflecting an eastward along-arc migration of the depocentres through Mid- to Late Cretaceous times (Miyata 1990; Yamakita & Ito 1999). Following the Ryoke metamorphism, and likely in continuity with it, the two accretionary complexes were subjected to voluminous and widespread acidic to intermediate magmatic events, which lasted from the Early Cretaceous to the end of the Palaeogene, with a peak during Late Cretaceous, and which led to the emplacement of two main groups of rocks: volcanic and volcanoclastic deposits locally including sedimentary strata, and plutonic to hypovolcanic intrusive rocks (Murakami 1975; Murakami etal 1989). Whole-rock, amphibole and muscovite K-Ar and zircon fission track datings of the volcano-sedimentary cover yielded ages between 115 and 67 Ma (Nishimura & Imaoka 1995). The Cretaceous plutons are distributed near the Seto Inland Sea (so-called San-yo volcanism). Their biotite K-Ar ages range from 115 to 56 Ma, while their whole-rock or biotite Rb-Sr ages range from 103 to 56 Ma (Nakajima et al 1990; Nishimura & Imaoka 1995). There is a clear along-arc lateral variation of K-Ar and Rb-Sr ages, consisting of an eastward younging from 105 to 80 Ma ages in the western Chugoku region to 75-56 Ma ages near central Japan (Nakajima et al. 1990). In contrast, Palaeogene volcano-plutonic complexes, with radiometric ages between 31 and 39 Ma (biotite Rb-Sr, whole-rock K-Ar and zircon fission track methods; Imaoka et al. 1988; Nishimura & Imaoka 1995) are present along the Japan Sea side (so-called San-in plutonism). Eocene to Miocene volcanic or sedimentary deposits locally cover Palaeogene volcano-plutonic complexes and older rocks. Lastly, a SW-NE alignment of Quaternary monogenetic volcanoes defines the present volcanic front linked with the subduction of the Philippine Sea plate (Fig. 1; Research Group for Active Faults of Japan 1991; Kamata 1998). Interpretations of the across-arc structural arrangement are based on the processes of forearc frontal sedimentary accretion and intra-arc igneous activity, both processes being regarded as resulting from more or less continuous subduction of oceanic plates under the Japanese margin through Palaeozoic to Cenozoic times (Uyeda & Miyashiro 1974; Taira et al. 1982, 1983; Maruyama & Seno 1986; Mackenzie et al. 1987; Taira & Ogawa 1988; Agar 1990;

145

DiTullio & Byrne 1990; Hibbard & Karig 1990; Isozaki et al 1990; Kimura & Mukai 1991; Kimura 1997; Maruyama et al. 1997). Another group of models calls for collisions between the Japanese margin and continental microblocks or arc fragments carried by the subducting plates (Faure et al. 1986; Maruyama & Seno 1986; Charvet et al. 1990; Otsuki 1992; Stein et al. 1994). A third group of models proposes to account for the along-arc diachronous emplacement of Cretaceous to Palaeogene granitic intrusions by the subduction of an active mid-oceanic ridge beneath the Japanese arc at that time (Kinoshita & Ito 1986; Nakajima et al. 1990; Kiminami et al. 1994; Kinoshita 1995; Brown 1998).

Geological outline of the Western Chugoku fault system Regional faults in the Outer Zone are predominantly thrust faults striking more or less parallel to the trend of the arc and dipping shallowly to moderately to the north, NW or NE, depending on the fault strikes. Steeply dipping or vertical strike-slip faults remain scarce (Murata 1990). In contrast and in addition to the moderately dipping thrust faults inherited from the Jurassic or older orogenies, the Inner Zone is crosscut by a system of vertical or steeply dipping strikeslip faults. A general discussion of this fault system is provided by Kanaori (1990) and Kanaori et al. (1990). In the present chapter, we examine in detail the characteristics of the strikeslip faults in the western part of the Chugoku region, in the vicinity of the cities of Hiroshima and Yamaguchi, where they form what we refer to as the Western Chugoku fault system. The vertical to steeply dipping strike-slip faults can readily be seen on satellite images or on aerial photographs (Research Group for Active Faults of Japan 1980, 1991; Kanaori 1990; Kanaori et al. 1990). They are also represented on geological maps where there is a mappable offset associated with the faults. A particular area mapped by Yamada et al. (1986) clarifies the relationships between PreCretaceous low-angle faults, Cretaceous intrusions and post-intrusion strike-slip faults (Fig. 3). In this area, Permian units (Nishiki Group, hanging wall) are separated from Jurassic units (Kuga Group, footwall) by a major PreCretaceous gently dipping thrust fault. This thrust fault is sealed by Late Cretaceous granitic and granodioritic intrusions. Both the fault and the intrusions are in turn offset by a steeply dipping NE-SW strike-slip fault. The net

146

O. FABBRIErAL.

Fig. 3. Geological map of the Yoshiwa-Kake area (modified after Yamada et al. 1986) showing the relationships between low-angle thrust faults and high-angle strike-slip faults, two types of faults characterizing the western Chugoku region. Also shown are the Hikimi and Kake localities where the fault-slip data of Figure 11 were obtained.

amount of left-lateral offset along the Pre-Cretaceous fault is about 2.5 km. Southern boundary of the Western Chugoku fault system The southern limit of the Western Chugoku fault system is clearly the MTL. Strictly speaking, the MTL is the trace of a major fault which has been active since Cretaceous to Palaeocene times (Ichikawa 1980). MTL-related mylonites are common on the Ryoke side where they derive from granitoids (Kara et al. 1980; Takagi

1986). Cataclasites and gouges developed from Sanbagawa schists or Ryoke gneisses, granitoids and mylonites testify to shear motion in brittle conditions (Wibberley & Shimamoto 2003). Because of the linear trace and generally strong dips of the outcrop-scale fault planes and of the mylonitic foliation, the MTL has long been considered as a vertical fault, along which motion was mainly of strike-slip type. However, detailed mapping revealed that MTL-related mylonites locally show a flat or gently dipping foliation and that, in other places, the brittle faults separating the Sanbagawa schists from the

DISTRIBUTED STRIKE-SLIP FAULTING IN JAPAN

Ryoke rocks dip gently (Ohtomo 1993; Yamamoto 1994; Sakakibara 1995). These observations suggest that an early, gently dipping, proto-MTL was later cut by a steep-dipping MTL and were confirmed by recent geophysical studies revealing that the near-surf ace fault dip tends to decrease with depth to less than 30° at about 5 km depth (Yoshikawa et al 1992; Yusa et al 1992; Ito et al 1996). Close to the ISTL (Fig. 1), where the MTL has an approximately north-south direction, Dallmeyer & Takasu (1992) performed a systematic 40Ar/39Ar study along a transect from massive granite of the Ryoke belt to mylonitic equivalents close to the MTL. The least affected Ryoke granite sample yielded a plateau age of 69.9 +1.0 Ma that is interpreted as a magmatic cooling age. Two whole-rock protomylonite samples yielded plateau ages of 61.1 + 1.0 and 63.0 + 0.8 Ma, interpreted by the authors as dating the rapid cooling following the development of the protomylonitic fabric. These ages are similar to the 60-70 Ma ages obtained by the K-Ar dating of biotite from Ryoke mylonitic granites (Shibata & Takagi 1988). K-Ar datings performed on clay minerals from MTLrelated gouges yielded ages between 58 and 10 Ma (Takagi & Shibata 1992; Yamakita et al. 1995). Apatite fission track ages systematically decrease towards the MTL from about 55 Ma (at >5km from the fault zone) to 9-13 Ma closest to the fault zone (Tagami et al. 1988). Such young ages may reflect either hydrothermal alteration during brittle movement along the MTL (Takagi & Shibata 1992) or a Miocene thermal overprint due to shear heating (Tagami et al. 1988). Where the mylonitic foliation is vertical, the sense of shear is left-lateral (Ichikawa 1980; Takagi 1986). Where the foliation is flat-lying, the sense of shear indicates a combined leftlateral and reverse motion (Yamamoto 1994). Furthermore, syntectonic deposition of the Campanian-Maastrichtian Izumi Group testifies to a left-lateral activity of the MTL during the period considered (Miyata 1990; Yamakita & Ito 1999). Offset streams indicate that the presentday motion of the MTL is a right-lateral strikeslip one (Okada 1980, 1992; Research Group for Active Faults of Japan 1991; Tsutsumi et al. 1991). Between the Late Cretaceous left-lateral shear and the present right-lateral shear, there remains a long period during which the kinematics are poorly constrained. Structural analysis and K-Ar dating of gouges from a shear zone along the extension of the MTL in eastern Kyushu allowed Yamakita et al. (1995) to reconstruct an Oligocene (35-33 Ma)

147

reverse and left-lateral motion. On the basis of fault-slip data analysis, Fournier et al. (1995) documented a Miocene normal motion along the MTL in Shikoku. With the exception of these localized observations, an integrative history of motion along the MTL is still lacking. Northern boundary of the Western Chugoku fault system The northward extension of the fault system is concealed by the Japan Sea. According to Kanaori (1990), the fault system abuts against a major fault running parallel to the coast. Yamamoto (1993) and Itoh et al. (2002) provide geophysical and morphological evidence for such a major WSW-ENE fault bordering the northern coast of the Chugoku region, the Southern Japan Sea fault zone (SJSFZ, Figs 1 & 2). This fault could be a branch of the Tsushima-Yangsan-Sikhote-Alin fault system (Natal'in et al. 1986; Yoon & Chough 1995; Fig. 1) which formed a continuous and straight fault system along the Japanese arc before the formation of the Japan Sea back-arc basin in Oligo-Miocene times (Faure & Lalevee 1987; Faure & Natal'in 1992; Yamakita & Otoh 2000). Faulted Pleistocene deposits and offset Holocene streams indicate that the Quaternary motion of the SJSFZ combines reverse and leftlateral strike-slip (Itoh et al. 2002). Based on an alignment of historical and modern shallow (focal depth <70km) earthquakes in SW Japan, Gutscher & Lallemand (1999) proposed that a major strike-slip fault, the North Chugoku shear zone (NCSZ), was forming along the northern (Japan Sea) coast of the Chugoku region. The inferred trace of the NCSZ runs about 30 km southwards of the SJSFZ proposed by Itoh et al. (2002). As admitted by Gutscher & Lallemand (1999), probably because of its recent formation, the NCSZ has not yet produced any substantial surface displacement and has still to be mapped. We will thus rely on the observations of Itoh et al. (2002), which are, to date, more documented. Geometry of the Western Chugoku fault system Unlike low-angle reverse faults bounding PreCretaceous thrust sheets, most faults of the Western Chugoku system appear on the surface as rectilinear valleys or aligned depressions, and therefore dip steeply. They are readily mapped with the help of satellite images or aerial photographs. Figure 4 is a lineament map

Fig. 4. The Western Chugoku fault system mapped from a 1976 Landsat image (scale is about 1/700 000). C, cloudy area; H, Hiroshima City; M, Masuda City; Y, Yamaguchi City; MTL, Median Tectonic Line. Also represented are the areas covered by the maps of Figures 5 and 6.

DISTRIBUTED STRIKE-SLIP FAULTING IN JAPAN

149

Fig. 5. Detailed lineament map of the Yamaguchi area based on 1967-68 aerial photographs of the Geographical Institute of Japan (scale is about 1 /20 000) with senses of motion where established. Quaternary to present sense of motion are deduced from offset streams (locality (1)), shear sense indicators preserved in clayey gouges (locality (2)) or earthquake focal mechanisms (locality (3)). The star indicates the epicentre of the 25 June 1997 Mw 5.8 northern Yamaguchi Prefecture earthquake (event 2 of Figure 9). Locality SASA-01 is the site of the fault-slip data displayed on Figure 10.

drawn from a Landsat image. More detailed maps of two sub-areas, located by the rectangles on Figure 4 and given on Figures 5 and 6, were obtained by tracing lineaments on aerial photographs taken by the Geographical Institute of Japan. Our field investigations have shown that most lineaments correspond to faults. The only exceptions are short NW-SE lineaments observed to the north or NW of Hiroshima (Fig. 6), which, rather than faults, are the surface expression of

joints in granitic rocks widespread in this area (Iwamura 2001). At both satellite and aerial photograph scales, the overall pattern of the Western Chugoku fault system is dominated by first-order faults (master faults), which can be followed in the topography for distances up to 140 km. Length-v.-direction histograms (Fig. 7) show that master faults strike N35°-N40°E in the Yamaguchi area, and N40°-N45°E in the Kake-Hikimi area, that is

150

O. FABBRI£7AL.

Fig. 6. Detailed lineament map of the Kake-Hikimi area (north of Hiroshima) based on 1967 aerial photographs of the Geographical Institute of Japan (scale is about 1/40 000).

25° to 30° counterclockwise to the boundary faults, namely the MIL and the SJSFZ. Second-order faults have shorter traces and are less clear on satellite images. They always abut against first-order faults (except west of Yamaguchi City, Fig. 4), and their directions are more scattered than those of the first-order faults. In particular, there is no clear peak on the lengthv.-direction histograms (Fig. 7). Two categories of faults can, however, be distinguished: (1) N10°-N35°E slightly oblique faults striking 15° to 25° counterclockwise from adjacent first-order faults, commonly showing curved traces, and in some cases connecting first-order faults between

them; and (2) N135°-N160°E highly oblique to perpendicular faults or fracture zones, always short and mostly developed around Hiroshima City (peak at 90° to the strike of the master faults; Fig. 7, Kake and Hikimi histograms), where they actually correspond to zones of joints in granitic plutons (Iwamura 2001). Field occurrence of the Western Chugoku system faults As a result of alluvial filling, the fault core zones remain poorly exposed. Where observable, they

DISTRIBUTED STRIKE-SLIP FAULTING IN JAPAN

151

Fig. 7. Lineament length-v.-direction in the Yamaguchi and Kake-Hikimi areas (percentages of cumulative lengths within 5° intervals). Total cumulative length is 1045 km for the Yamaguchi area and 712 km for the Hikimi-Kake area.

consist of several centimetres to several metres thickness of vertical layers of cataclasites, breccias and gouges inside which steeply dipping (>70°) slickenside surfaces bearing horizontal striations can be recognized. No mylonitic rocks have been observed. First-order fault core zones cannot be distinguished from secondorder fault core zones on the basis of faultrelated rocks or internal structure. The only noticeable difference lies in the size of the core zones: the first-order fault core zones are generally wider than the second-order ones. Figure 8 provides some information regarding the geometry of the slickenside surfaces found

in or near the fault core zones of the Yamaguchi area (Matsunaga 1997). The directions of the surfaces show a peak at N20°-N30°E and another peak at NO°-N10°E (Fig. 8a). Neither peak corresponds to peaks in the strikes of the regional faults (Fig. 7 a). In fact, most slickenside surfaces in core zones from the Yamaguchi area are oblique to the strikes of the host faults. What is more demonstrative is the distribution of slickenside surface dips (Fig. 8b) and that of striation rakes (Fig. 8c). Dips larger than 70° clearly predominate, as do rakes smaller than 20°. This confirms the basic strike-slip nature of the steeply dipping faults, at least in the Yamaguchi area.

152

O. FABBRIETAL.

Fig. 8. Geometry of the slickenside surfaces found in or near the fault core zones of the Yamaguchi area (Matsunaga 1997). Number of data: 603. (a) Directions of slickenside surfaces; (b) dips of slickenside surfaces; (c) rakes of striations borne by the slickenside surfaces.

Age of the Western Chugoku fault system Neither absolute age constraints on fault activity, for example by radiometric dating of fault-related rocks (cataclasite or gouge, since mylonites are lacking), nor relative ages, for example by dating a non-cataclastic dyke crosscutting a cataclastic belt, are available in the study area. The age of formation and motion history of the Western Chugoku fault system, and more generally of the Inner Zone fault system, are therefore poorly constrained. Some information can, however, be obtained from the displacement history of the boundary faults. As already mentioned, 40Ar/39Ar or K-Ar dating of MTL-related mylonitic rocks yielded 70-60 Ma ages, which are the ages of the ductile deformation (Shibata & Takagi 1988; Dallmeyer & Takasu 1992). These ages show that the MTL was active at that time and possibly also before. If the Pre-Quaternary displacement history along the SJSFZ is unknown, some chronological information is available for the nearby, and probably cogenetic, Sikhote-Alin fault system. Displacement along the SikhoteAlin fault system occurred between the Early Cretaceous and the Palaeogene, with a maximum during the Late Cretaceous, and was left-lateral (Natal'in et al 1986; Faure & Natal'in 1992). These chronological indicators show that the boundary faults were active at the end of the Cretaceous through to the beginning of the Tertiary, leaving open the possibility for the intervening Western Chugoku and more generally for the Inner Zone fault systems to have formed at that time. Formation of first-order N45°E faults in the Hiroshima region at Late Cretaceous times is further supported by the emplacement of

Late Cretaceous elongated granite porphyry dykes or stocks along the faults, some of which possibly show an en echelon arrangement (e.g. Yamada et al 1986; Fig. 3). Fault kinematics Geological maps of the western Chugoku area (Yamada et al 1986; Nishimura et al 1995) show that mappable markers such as ring dykes or contacts between pluton and country rock are left-laterally offset along first-order N45°E faults. The amount of apparent (cumulative) horizontal offset for these Cretaceous or older rocks ranges between 1.5 and 3.5 km (e.g. Fig. 3). No significant offset along second-order faults can be detected from examination of published maps. Beside cartographic indications, detailed kinematical analysis shows that some of the faults of the Western Chugoku fault system underwent at least one inversion of the sense of motion (Figs 4 to 6). In fact, as detailed in the next sections, around 2 Ma, first-order faults changed from leftlateral to right-lateral, whereas some secondorder faults experienced the opposite reactivation (from right-lateral to left-lateral). Plio-Quaternary to present-day kinematics Plio-Quaternary to present-day kinematic indicators are of three types: (1) offset topographic features, (2) earthquake focal mechanisms along with aftershock distributions, and (3) shear sense indicators observed in incohesive gouges in fault zone cores. In the study area (Fig. 4), offset topographic markers are found only along the first-order

DISTRIBUTED STRIKE-SLIP FAULTING IN JAPAN

153

Fig. 9. Focal spheres of some crustal earthquakes in the Chugoku region (compiled after Ando 1995; Fukuyama et al. 2000; Ohmi 2002).

faults where they testify to a right-lateral motion (Research Group for Active Faults of Japan 1991). To the east of the study area, offset topographic markers observed along second-order N135°E faults (highly oblique to perpendicular type) indicate a left-lateral motion (Research Group for Active Faults of Japan 1991; Inoue et al 2002). Focal mechanisms of upper crustal (0-20 km) events in the Chugoku region commonly display two vertical nodal planes, one about N45°N60°E and the other about N135°-150°E (Fig. 9; see also Gutscher & Lallemand 1999). At about 300 km to the east of the study area, the 25 January 1995 Mw 7.2 Kobe earthquake (Fig. 9, event 9) was generated by the rupture of a steeply dipping fault trending N50°N60°E (Ando 1995). The focal mechanism of the main shock and coseismic surface deformation clearly indicate a right-lateral slip (Ando 1995; Lin et al 1995; Lin & Uda 1996). Though smaller in magnitude, two earthquakes recently occurred in the upper crust of the western Chugoku area. The 25 June 1997

Mw 5.8 Northern Yamaguchi Prefecture earthquake (Fig. 9, events 2 and 3) was induced by a right-lateral slip along a well-expressed N45°Etrending fault, though the rupture did not reach the surface (Kanaori et al 1999; Fukuyama et al 2000). Conversely, just to the east of the study area, the 6 October 2000 Mw 6.5 Tottori earthquake (Fig. 9, event 4) was caused by a left-lateral slip along a fault trending about N145°E (Sagiya et al 2002). In this last case the fault that slipped is expressed at the surface not by a well-marked lineament but rather by a series of discontinuous fault segments trending N145°E along which streams and other topographic features are offset left-laterally (Inoue et al 2002). The Plio-Quaternary to present-day kinematics of the Western Chugoku fault system are compatible with the present-day stress field active throughout SW Japan, which is characterized by an east-west principal horizontal maximum stress axis crHmax and a north-south principal horizontal minimum stress axis OHmin (Ichikawa 1971; Huzita 1980; Tsukahara & Kobayashi 1991).

154

O. FABBRICTAL.

Fig. 10. Lower-hemisphere equal-area projection of the fault-slip data observed at locality SASA-01 (location on Fig. 5). Number of data: 53.

Pre-Plio-Quaternary kinematics Pre-Plio-Quaternary kinematics are not easily determined, mainly because of the complexity of damage zones and the difficulty of establishing a hierarchy among only partly exposed cataclastic belts and slickenside surfaces. The net left-lateral kilometric offset observed along first-order N45°E faults (Fig. 3; Yamada et al 1986; Nishimura et al. 1995) cannot be explained by the recent right-lateral motion described in the previous section. It implies that the recent motion was preceded by a left-lateral motion. As detailed below, this is in agreement with outcrop-scale kinematical criteria observed along slickenside surfaces, which are of three types: (1) offset geological markers (dykes, xenoliths, key beds, etc.), (2) secondary fibrous minerals on the slickenside surfaces, and (3) en echelon R-type fractures adjacent to the surfaces. The same criteria also indicate a right-lateral slip along N135°-N160°E secondary faults, in disagreement with the recent left-lateral sense of slip described above. In this case, too, we interpret this apparent discrepancy as simply reflecting a diachronism in the two stages of motion. Figure 10 displays fault-slip data from one outcrop located 15 km north of Yamaguchi (Fig. 5, locality SASA-01). Outcrop SASA-01 is located at a few metres from the junction between a first-order N50°E fault and an abutting second-order N155°E fault. The junction itself is

not exposed but its proximity is suggested by the anomalously high density of fractures and the intensity of hydrothermal alteration. Unambiguous kinematic indicators are associated with abundant slickenside surfaces. Two clusters of faults can be easily recognized within the dataset: N30°-N75°E left-lateral faults and N125°-N175°E right-lateral faults. A third cluster includes six planes striking between N170°E and N30°E. The first cluster of fault planes (N30°-N75°E) is parallel to the trend of the nearby first-order fault, and the slip senses are homogeneous, suggesting a left-lateral sense of slip along the N50°E fault. Similarly, the parallelism and slip sense homogeneity of the outcrop-scale faults suggest a right-lateral sense of slip along the N155°E fault. Further insights can be obtained from two fault exposures near Hiroshima (Fig. 3, Kake and Hikimi outcrops). Along fault planes of a given trend, both left-lateral and right-lateral slip senses are observed (Fig. 11). This is particularly obvious at the Kake locality for NE-SW faults which parallel a nearby first-order fault: shear bands within gouge zones clearly demonstrate a right-lateral motion, whereas quartz fibres on the planes bounding the gouge zones indicate a left-lateral motion. Opposite slip senses are also observed on perpendicular NW-SE faults at Hikimi, but the data are less numerous. Given the reliability of the slip sense criteria, we are forced to admit that these faults have recorded an inversion of the strike-slip motion. The same observations and consequent interpretation were made by Kanaori (1999) along a core zone of a first-order lineament exposed NE of Yamaguchi (Fig. 5, locality (2)): the arrangement of R-type antithetic fractures branching from slickenside surfaces encompassing a clayey gouge zone indicates a left-lateral sense of slip whereas P-like schistosity within the clayey gouge zone indicates a right-lateral sense of slip. To date, the slip senses observed along outcrop-scale slickenside surfaces exposed near the N10°-N40°E slightly oblique second-order faults are left-lateral (Matsunaga 1997; Iwamura 2001). No clear inversion of the sense of movement could be established. In summary, both right-lateral and left-lateral kinematics are recorded along N45°E first-order (master) faults and along N135°-N160°E (highly oblique to perpendicular) second-order faults. The best way to reconcile such a complexity is to call for a diachroneity in the motion history of the faults, at all scales, that is from the scale of the outcrop to that of the study area. Such a polyphase history is summarized on Figure 12.

DISTRIBUTED STRIKE-SLIP FAULTING IN JAPAN

155

Fig. 11. Interpretation of fault-slip data from the Kake and Hikimi outcrops (location on Fig. 3). For both sites, the data are separated into two subsets corresponding to two contrasted stress tensors, type 1 and type 2. The projections of the principal stress axes 0-b a2 and cr3 are symbolized by five-, four- and three-pointed stars, respectively.

Age of the recent inversion of the sense of motion The age of the tectonic reactivation cannot, at present, be precisely determined. However, it can be estimated from consistent geological observations from central Japan where the IzuBonin arc (Philippine Sea plate) has been colliding with the Japanese arc (Eurasian plate) since at least 10 Ma ago (Fig. 1; Amano 1989; Niitsuma 1989). The direction of relative convergence between the Philippine Sea plate and the Eurasia plate is NW-SE (Seno et al 1993). In Central Japan, where the Bonin arc is currently colliding with the Japan arc, the NW-SE convergence is responsible for a fan-shaped crHmax trajectory radiating north and NW of the Izu peninsula (Fig. 1). Based on an analysis of

fault-slip data at the rear of the Izu collision zone, Angelier & Huchon (1987) were able to reconstruct a Pre-Quaternary o~Hmax trajectory linked with the Pliocene collision between the Izu-Bonin arc and the Japan arc. The more northerly o~Hmax trajectory that was reconstructed suggests that the direction of relative convergence between the two plates was closer to north-south than it is today. The stratigraphic ages of the youngest strata having recorded the more northerly crHmax trajectories are estimated as between 1 and 3 Ma, depending on the location with respect to the collision zone and the uncertainties of the stratigraphic ages. The present-day stress field started to be active from some time between 3 and 1 Ma. The northsouth to NW-SE shift of the direction of relative convergence has also been recognized by the

156

O. FABBRICTAL.

Fig. 12. Polyphase kinematic history of the Western Chugoku fault system. The crHmax and o^Hmin stress components are deduced from fault-slip data inversion (pre-2 Ma stage) and from earthquake focal mechanisms (post-2 Ma stage).

same method (reconstruction of 0Hmax from fault-slip data) in Taiwan, which is another area where the Philippine Sea and Eurasia plates collide (Angelier et al 1986). In that area, the change of the convergence direction is not accurately dated, but is estimated to have occurred between 4 and 1 Ma. As a whole, 2 + 1 Ma is a possible date for the recent tectonic inversion recorded in the study area.

Similarities with analogue models and genetic implications The Western Chugoku fault system is a typical example of a zone of distributed shear deformation characterized by: (1) a 60-90 km wide zone of conjugate discrete strike-slip faults, (2) the lack of any single prominent fault, and (3) a weak amount of left-lateral cumulative offset along the first-order faults (a few kilometres offset for 120 km or longer fault traces). These characteristics show that the studied fault system is developed not above a unique deep wrench basement fault but rather above a wide zone with transcurrent deformation. Additional information about the geometry at depth can be deduced from a comparison with the fault patterns obtained by experimental analogue modelling. Indeed, the pattern of the Western Chugoku fault system displays a striking similarity with patterns obtained by Schreurs (1994) during experiments of simple shear deformation distributed at the base of a brittle layer analogue over its entire width (basal distributed flow). The overall sense of shear in Schreurs' (1994) experiments was right-lateral. In order to allow a comparison with the Western

Chugoku fault system, Figure 13 displays a mirror image of the fault pattern obtained by Schreurs (1994). The overall sense of shear is left-lateral. The angular distribution of the obtained faults is given on Figure 14. The comparison between Figures 13 and 14 and Figures 4, 5, 6 and 7 shows the following similarities: 1

regular spacing of the first-order faults, whose motion is synthetic of the overall left-lateral shear; 2 same 25-30° acute angle between the firstorder faults and the boundary faults; 3 existence of slightly oblique secondary faults (RL on Fig. 13) linking first-order faults between them and possibly having sigmoidal S-like traces; 4 clear abutting relationships between firstorder faults and highly oblique to perpendicular second-order faults, indicating that the first-order faults developed first. Differences are also noticeable: 1

numerous highly oblique to perpendicular second-order faults are observed in the Western Chugoku fault system, but not in the modelling; 2 slightly oblique second-order faults of the model (R£ on Fig. 13) are clearly antithetic (right-lateral in the mirror image of Fig. 13), whereas the slightly oblique second-order faults are predominantly synthetic (leftlateral) in the natural example; 3 in the most evolved stage of the analogue modelling (Fig. 13D), cross-faults (RL) experience a small dip-slip component that is not observed to date in the study area.

DISTRIBUTED STRIKE-SLIP FAULTING IN JAPAN

157

Fig. 13. Fault pattern obtained by analogue modelling of basal distributed left-lateral shear. The pattern depicted here is a mirror image of that obtained by Schreurs (1994) in an overall right-lateral shear. R and RL are synthetic left-lateral faults. RL are antithetic right-lateral faults. Faults with ticks show some dip-slip component.

The differences can be explained by inherent features of the Western Chugoku fault system. Firstly, in the natural case of SW Japan, the strike-slip faults are developed on a lithologically and Theologically strongly heterogeneous substratum made up of superposed accretionary complexes that contain igneous intrusions, in contrast to the homogeneous analogue model of Schreurs (1994). Secondly, comparison between analogue and natural models should

not be driven too far because of scaling factors and the fact that such models are often dry, that is there is no pore fluid pressure. Thirdly, as already stated, the abundance of perpendicular second-order faults in the Chugoku area may reflect pre-existing pervasive jointing in Cretaceous granitic rocks as well as in Cretaceous or Palaeogene volcanic rocks (mostly ignimbrites). Since the analogue modelling does not include fractures in its initial state,

Fig. 14. Fault length-v.-direction for step D of the analogue modelling of basal distributed left-lateral shear (Fig. 13D; percentages of cumulative lengths within 5° intervals).

158

O. FABBRI ETAL

the effects of pre-existing fractures cannot be checked. The second difference between the Chugoku natural example and the analogue models lies in the kinematics of the slightly oblique secondorder faults, which seem to be left-lateral (synthetic) in the natural example but are right-lateral (antithetic) in the experiment. This difference cannot be satisfactorily explained without additional kinematic evidence concerning the slightly oblique faults in the study area. The same can be said for the apparently missing dip-slip components in the study area. Despite these differences, the geometrical similarities between the natural and the experimental fault patterns suggest a similar genesis in both cases through the means of a decoupling zone at the base of the brittle layer, which would allow a homogeneous distribution of the shear deformation. Given the fact that the southern limit of the Western Chugoku fault system, that is the MTL, seems to flatten at or below 5 km in depth, it is tempting to consider it as the decoupling/coupling plane at the time of formation of the fault system, that is in Late Cretaceous to Palaeocene times. This scenario depends of course on two unknowns: (1) the geometry of the MTL at depth at the time of formation of the fault system, and (2) the original geometry of the fault system, since the presently observable geometry has been modified, at least by the Plio-Quaternary reactivation described above and also by Mid-Tertiary vertical-axis rotations revealed by palaeomagnetic measurements and related to the opening of the Japan Sea, as detailed in the next section. Vertical-axis clockwise rotations and strike-slip faulting Palaeomagnetic measurements carried out on Cretaceous to Miocene rocks from the western Chugoku region have revealed a mean 50° clockwise deflection of the expected declination (Fig. 2), thus suggesting a clockwise rotation of the area around a vertical axis (Otofuji & Matsuda 1983, 1984, 1987; Otofuji el al 1985, 1991). Rocks younger than c. 14 Ma do not present deflected declinations. Their emplacement post-dates the vertical-axis rotations. According to Otofuji et al (1991), 80% of the 50° clockwise deflections occurred between 16 and 14 Ma, while the remaining 20% would reflect older rotations, possibly as old as Cretaceous or Palaeogene. However, the 16-14 Ma timespan should not be regarded as definitive, since the ages of rotated/non-rotated rocks are averaged radiometric ages having non-negligible

error bars (Jolivet et al. 1995). Whatever the accuracy of the ages, it is clear that most of the clockwise rotations occurred some time during the Early to Mid-Miocene, that is during the opening of the Japan Sea back-arc basin, which formed between 30 and 12 Ma (Tamaki et al 1992). The clockwise rotations were initially interpreted as the consequence of a rigid rotation (without any internal deformation) of the SW Japan arc (Otofuji et al 1985, 1991). This interpretation did not take into account the block structure of SW Japan, as initially described by Kanaori (1990), and the possibility of block rotation during shear (bookshelf or domino-type models; Mandl 1987). In their model of the Japan Sea opening, Jolivet et al (1995) consider that 30° out of the 50° deflection reflects the rotation of crustal blocks, the remaining being due to the overall net rotation of the arc. In fact, the respective amounts of rotation due either to internal block deformation or to rigid arc rotation can be determined as follows. On the basis of pre-Japan Sea opening reconstructions, which are independent from palaeomagnetic constraints, the MTL was parallel to the Yangsan-Tsushima-Sikhote-Alin faults, and it had an original strike of about N20°E (Faure & Lalevee 1987; Faure & Natal'in 1992; Yamakita & Otoh 2000). At the longitude of the western Chugoku region, the MTL presently strikes about N65°E. If the reconstructions are reliable, it implies that the MTL rotated clockwise about 45° during the opening of the Japan Sea. Since the total amount of rotation estimated by palaeomagnetic investigations is about 50°, the clockwise rotation due to internal deformation is about 5°. Given the trend of the intervening first-order faults, a clockwise rotation coeval with left-lateral strike-slip along the boundary faults (bookshelf mechanism) is possible only if the deformation is transpressional, that is if there is some amount of shortening perpendicular to the boundary faults (Mattel et al 1988). If shortening is not possible, either there is extension across the boundary faults (transtensional deformation) and, in this case, blocks would rotate counterclockwise, which is not supported by the data, or there is no change in the width of the system and, in this case, internal rotation is impossible. The proposed model will therefore be based on a clockwise block rotation limited to about 5°, possibly taking place as early as the Late Cretaceous, during formation of the Inner Zone fault system. The 5° figure should not be regarded as a definite and accurate value, but rather as an indication of the relative importance of internal v. overall rotations.

DISTRIBUTED STRIKE-SLIP FAULTING IN JAPAN

Model of formation and evolution of the Western Chugoku fault system Based on geometrical and kinematical similarities with fault patterns obtained by analogue modelling, we propose that the Western Chugoku fault system formed by distributed basal shear of the brittle crust located above the horizontal part of the proto-MTL, and between the proto-Sikhote-Alin fault and the vertical part of the proto-MTL. The model includes three main steps: 1

(Late Cretaceous-Palaeocene). Distributed left-lateral basal shear above the flat-lying proto-MTL induces the formation of firstorder left-lateral faults and second-order right-lateral faults (Fig. 15a). Vertical-axis block rotation, if any, should be clockwise (thus requiring a component of across-arc shortening, Fig. 15b) and should be limited to a small amount (about 5°). 2 (Early to Mid-Miocene). During the opening of the Japan Sea and the concurrent clockwise rotation of the SW Japan arc, leftlateral slip along the first-order faults resumes and may be accompanied by some limited amount (about 5°) of clockwise rotation of internal blocks following transpressional left-lateral shear along the boundary faults (Fig. 15b). 3 (Plio-Quaternary). Major plate convergence reorganization around 2 Ma leads to inversion of the sense of motion along boundary

159

faults and also along block-bounding faults (Fig. 12). In accordance with the Schreurs' (1994) model, the overall geometry of the Western Chugoku fault system requires that the general sense of shear is left-lateral in Late Cretaceous-Palaeocene times (step 1; Fig. 15a). It implies that the greatest principal stress component crHmax, likely equal to &i (the vertical principal stress axis being the intermediate principal axis 02), trends at about 30-45° counterclockwise from the faults bounding the system. Such a boundary condition can be achieved by oblique convergence between the subducting oceanic plates and the SW Japan arc. Indeed, several palaeogeographical reconstructions suggest that oblique subduction was taking place in the Japanese convergent margin in Late Cretaceous times (Kiminami et al 1994; Kimura 1997; Brown 1998). The model proposed here differs from that proposed by Kanaori (1990) on the following points. Firstly, in the model of Kanaori (1990), shear of the Inner Zone of SW Japan is side-driven, that is stresses are transmitted through two vertical faults, the proto-MTL and the proto-SikhoteAlin fault. In the model proposed here, shear is bottom-driven and stresses are transmitted through the whole flat-lying part of the protoMTL viewed as a partial attachment or clutch zone (Tikoff et al. 2002). Secondly, the firstorder faults, which are the first intervening faults to appear, strike c. 30° counterclockwise

Fig. 15. Model of formation and subsequent evolution of the Western Chugoku fault system, (a) Step 1 (Late Cretaceous-Palaeocene): Distributed left-lateral basal shear along boundary faults (proto-MTL) induces the formation of first-order left-lateral faults and second-order right-lateral faults in the overlying crustal domain, (b) Step 2 (Early to Mid-Miocene): Regional transpressional left-lateral shear caused by general clockwise rotation of the SW Japan arc during the Japan Sea opening is reponsible for resumption of strike-slip faulting (left-lateral slip along first-order faults, right-lateral slip along second-order faults), and limited (about 5°) vertical-axis clockwise rotation of internal blocks. Transpression is also responsible for a decrease of the width of the entire shear zone (Martel et al. 1988).

160

O. FABBRICTAL.

to the boundary faults (Fig. 15a) whereas they strike about 75° counterclockwise in the Kanaori model. Thirdly, rotation of internal blocks is limited to about 5° here (estimation based on a comparison of the pre- and postJapan Sea opening strikes of the MTL), whereas it reaches 35° in the Kanaori model. Yamamoto (1994) explained the formation of MTL-related flat-lying mylonites by basal shear along a horizontal zone located in the uppermost part of the ductile crust. At the inferred time of formation of the Western Chugoku (and Inner Zone) fault system, that is in Late CretaceousPalaeocene times, the study area was the site of an intense magmatic activity, possibly linked with the subduction of an active mid-oceanic ridge (see above). With a high geothermal gradient of 50 °C km"1, the depth of the uppermost part of the ductile zone would be around 7 km, which is an estimate of the depth of the protoMTL at the time of its formation. Following uplift and crustal cooling since the Cretaceous, the proto-MTL went through the brittle-ductile transition zone and reached the brittle domain at a depth of about 5 km, which corresponds to the estimated present depth of the flat part of the MTL (Ito et al 1996).

Schreurs (1994) does not explain all the geometrical and kinematical characteristics of the Inner Zone fault system. In particular, the sense of slip along second-order faults striking 45°65° counterclockwise of the boundary faults (slightly oblique second-order faults in our nomenclature; RL of Schreurs 1994) and expected to be antithetic (right-lateral) appears to be synthetic (left-lateral) in the field. This discrepancy may result from the fact that the Inner Zone system has suffered from an inversion of the sense of slip following a Plio-Quaternary reorganization of external stresses. Ancient kinematics are therefore more difficult to unravel. More generally, the long-lived natural system described here cannot be as simple as that obtained experimentally. Future work should aim at a more accurate mapping of block boundaries, a difficult task given the scarcity of exposures (not to mention areas covered by the sea) and the stratigraphical uncertainties, which prevent correlations from one fault side to the other. Beside mapping efforts, seismic reflection investigations across the entire SW Japan arc could provide additional constraints for the hypothetical model proposed here.

Conclusion The distributed characteristics of the Western Chugoku fault system, and more generally of the Inner Zone of SW Japan, pose the question of the downward extension of these structures. This question is important, since only a clear view of the infrastructure can lead to a better understanding of the mechanical and seismotectonic behaviour of the crust of the SW Japan arc. Based on geometrical and kinematical similarities with experimental analogue modelling, we explain the distributed strike-slip faulting of the Western Chugoku fault system and more generally of the Inner Zone of SW Japan by basal shear along the flat part of the proto-MTL viewed as a partial attachment zone (Tikoff et al. 2002). The apparent synchroneity between (1) the formation of the Inner Zone distributed strike-slip fault system and (2) possible subduction of a mid-oceanic ridge between SW Japan (Kinoshita & Ito 1986; Nakajima et al 1990; Kiminami et al. 1994; Kinoshita 1995; Brown 1998) may not be fortuitous. Indeed, the heat provided by the ridge subduction is likely to have softened the base of the brittle crust, thus allowing a better lateral distribution of the basal stresses to the overlying brittle crust. The model of distributed basal shear as deduced from the analogue modelling of

Fieldwork was supported by the French Ministry of Foreign Affairs (Ministere des Affaires Etrangeres) and the Japanese Ministry of Education. Serge Andre is acknowledged for drawing the figures. Careful reviews by J. Grocott, G. K. Taylor and K. De Jong helped to improve the manuscript.

References AGAR, S.M. 1990. The interaction of fluid processes and progressive deformation during shallow level accretion: examples from the Shimanto Belt of SW Japan. Journal of Geophysical Research, 95, 9133-9147. AGAR, S.M., CLIFF, R.A., DUDDY, I.R. & REX, D.C. 1989. Accretion and uplift in the Shimanto Belt, SW Japan. Journal of the Geological Society of London, 146, 893-896. AMANO, K. 1989. Multiple collision tectonics of the south Fossa Magna in central Japan. Modern Geology, 14,315-329. ANDO, M. 1995. Triggering, foreshocks, main shock and aftershocks of the 1995 Hyogo-ken Nanbu earthquake. Gekkan Chikyu (The Earth Monthly), 13, 18-29. ANGELIER, J. & HUCHON, P. 1987. Tectonic records of convergence changes in a collision area: the Boso and Miura peninsulas, central Japan. Earth and Planetary Science Letters, 81, 397-408. ANGELIER, J., BARRIER, E. & CHU, H.T. 1986. Plate collision and paleostress trajectories in a

DISTRIBUTED STRIKE-SLIP FAULTING IN JAPAN fold-thrust belt: the foothills of Taiwan. Tectonophysics, 125, 161-178. Aw AN, M.A. & KIMURA, G. 1996. Thermal structure and uplift of the Cretaceous Shimanto Belt, Kii Peninsula, Southwest Japan. The Island Arc, 5, 69-88. BROCHER, T.M., MCCARTHY, J., et al 1994. Seismic evidence for a lower-crustal detachment beneath San Francisco Bay, California. Science, 265, 1436-1439. BROWN, M. 1998. Unpairing metamorphic belts: P-T paths and a tectonic model for the Ryoke Belt, southwest Japan. Journal of Metamorphic Geology, 163, 3-22. BROWN, M., DIAZ, F. & GROCOTT, J. 1993. Displacement history of the Atacama fault system 25°00/S27°00/S, northern Chile. Geological Society of America Bulletin, 105, 1165-1174. BURGMANN, R. 1997. Active detachment faulting in the San Francisco Bay area? Geology, 25, 1135-1138. CARIDROIT, M., FAURE, M. & CHARVET, J. 1986. Nouvelles donnees stratigraphiques et structurales sur le Paleozoi'que superieur des zones internes du Japon sud-ouest. Un essai sur 1'orogenese permienne. Bulletin de la Societe Geologique de France, 8, 683-691. CHARVET, J., FAURE, M., FABBRI, O., CLUZEL, D. & LAPIERRE, H. 1990. Accretion and collision during East-Asiatic margin building - A new insight on the peri-Pacific orogenies. In: WILEY, T.J., HOWELL, D.G. & WONG, F.L. (eds) Terrane Analysis of China and Pacific Rim. Circum-Pacific Council for Energy and Mineral Resources, Houston (Texas), Earth Sciences Series, 13, 161-191. DALLMEYER, R.D. & TAKASU, A. 1992. Middle Paleocene terrane juxtaposition along the Median Tectonic Line, southwest Japan: evidence from 40 Ar/39Ar mineral ages. Tectonophysics, 200, 281-297. DALLMEYER, R.D., TAKASU, A. & YAMAGUCHI, K. 1995. Mesozoic tectonothermal development of the Sanbagawa, Mikabu and Chichibu belts, southwest Japan: evidence from 40Ar/39Ar whole-rock phyllite ages. Journal of Metamorphic Geology, 13,271-286. DICKINSON, W.R. 1996. Kinematics of Transnational tectonism in the California Transverse Ranges and Its Contribution to Cumulative Slip Along the San Andreas Transform Fault System. Geological Society of America Special Paper 305. DiTuLLio, L.D. & BYRNE, T.B. 1990. Deformation in the shallow levels of an accretionary prism: the Eocene Shimanto belt of southwest Japan. Geological Society of America Bulletin, 102, 1420-1438. FAURE, M. & CHARVET, J. 1987. Late Permian/early Triassic orogeny in Japan: piling up of nappes, transverse lineation and continental subduction of the Honshu block. Earth and Planetary Science Letters, 84, 295-308. FAURE, M. & LALEVEE, F. 1987. Bent structural trends of Japan: flexural slip folding related to the Neogene opening of the Sea of Japan. Geology, 15, 49-52.

161

FAURE, M. & NATAL'IN, B. 1992. The geodynamic evolution of the eastern Eurasian margin in Mesozoic times. Tectonophysics, 208, 397-411. FAURE, M., CARIDROIT, M. & CHARVET, J. 1986. The Late Jurassic oblique collisional orogen of SW Japan. New structural data and synthesis. Tectonics, 5, 1089-1114. FOURNIER, M., JOLIVET, L. & FABBRI, O. 1995.

Neogene stress fields in SW Japan and mechanism of deformation during the Sea of Japan opening. Journal of Geophysical Research, 100, 2429524314. FUKUYAMA, E., ISHIDA, M., et al. 2000. NIED Seismic Moment Tensor Catalogue January-December 1997. Technical Note of the National Research Institute for Earth Science and Disaster Prevention, 205. GEOLOGICAL SURVEY OF JAPAN. 19920. Geological Map of Japan and Adjoining Areas, 1:4000 000. GEOLOGICAL SURVEY OF JAPAN. I992b. Geological Map of Japan, (3rd edn) 1:1000 000. GROCOTT, J. & TAYLOR, G.K. 2002. Magmatic arc fault systems, deformation partitioning and emplacement of granitic complexes in the coastal Cordillera, north Chilean Andes (25°30/S to 27°00'S). Journal of the Geological Society, London, 159, 425-442. GUTSCHER, M.A. & LALLEMAND, S.E. 1999. Birth of a major strike-slip fault in SW Japan. Terra Nova, 11, 203-209. HADA, S. & KURIMOTO, C. 1990. Northern Chichibu terrane. In: ICHIKAWA, K., MIZUTANI, S., HARA, I., HADA, S. & YAO, A. (eds) Pre-Cretaceous Terranes of Japan. Publication of IGCP Project 224, 165-183. HARA, L, SHYOJI, K., SAKURAI, S. & HIDE, K. 1980. Origin of the Median Tectonic Line and its initial shape. In: ICHIKAWA, K. (ed.) Median Tectonic Line of Southwest Japan. The Memoirs of the Geological Society of Japan, 18, 27-49. HARA, I. SHIODA, T. HIDE, K., OKAMOTO, K., TAKEDA, K., HAYASAKA, Y. & SAKURAI, Y. 1990. Nappe structure of the Sanbagawa belt. Journal of Metamorphic Geology, 8, 441-456. HARA, L, SHIOTA, T., et al. 1992. Tectonic evolution of the Sanbagawa schists and its implications in convergent margin processes. Journal of Science of the Hiroshima University, 9, 495-595. HEMPTON, M.R. & NEHER, K. 1986. Experimental fracture, strain and subsidence patterns over en echelon strike-slip faults: implications for the structural evolution of pull-apart basins. Journal of Structural Geology, 8, 597-605. HIBBARD, J.P. & KARIG, D.E. 1987. Sheath-like folds and progressive fold deformation in Tertiary sedimentary rocks from the Shimanto accretionary complex, Japan. Journal of Structural Geology, 7, 845-857. HIBBARD, J.P. & KARIG, D.E. 1990. Structural and magmatic responses to spreading ridge subduction: an example from Southwest Japan. Tectonics, 9, 207-230. HUZITA, K. 1980. Role of the Median Tectonic Line in the Quaternary tectonics of the Japanese Islands.

162

O. FABBRICTAL.

In: ICHIKAWA, K. (ed.) Median Tectonic Line of Southwest Japan. Memoirs of the Geological Society of Japan, 18, 129-153. ICHIKAWA, K. 1980. Geohistory of the Median Tectonic Line of Southwest Japan. In: ICHIKAWA, K. (ed.) Median Tectonic Line of Southwest Japan. Memoirs of the Geological Society of Japan, 18, 5-24. ICHIKAWA, K., MIZUTANI, S., KARA, I., HADA, S. & YAO, A. (eds) 1990. Pre-Cretaceous Terranes of Japan. Publication of IGCP Project 224. ICHIKAWA, M. 1971. Reanalysis of mechanisms of earthquakes which occurred in and near Japan, and statistical studies on the nodal plane solutions obtained. 1926-1968. Geophysical Magazine, 35, 207-274. IMAOKA, T., MURAKAMI, N., MATSUMOTO, T. & YAMASAKI, H. 1988. Paleogene cauldrons in the western San-in district, Southwest Japan. Journal of the Faculty of Liberal Arts, Yamaguchi University, 22, 41-75. INOUE, D., MIYAKOSHI, K., UETA, K., MIYAWAKI, A. & MATSUURA, K. 2002. Active fault study in the 2000 Tottori-ken Seibu earthquake area. Journal of the Seismological Society of Japan, 54, 557-573. ISOZAKI, Y. 1996. Anatomy and genesis of a subdue tion-related orogen: a new view of geotectonic subdivision and evolution of the Japanese Islands. The Island Arc, 5, 289-320. ISOZAKI, Y. & ITAYA, T. 1991. Pre-Jurassic klippe in northern Chichibu belt in west-central Shikoku, southwest Japan: Kurosegawa terrane as a tectonic outlier of the pre-Jurassic rocks of the Inner Zone. Journal of the Geological Society of Japan, 97, 431-450. ISOZAKI, Y., MARUYAMA, S. & FURUOKA, F. 1990. Accreted oceanic materials in Japan. Tectonophysics, 181, 179-206. ITO, T., IKAWA, T., YAMAKITA, S. & MAEDA, T. 1996. Gently dipping Median Tectonic Line (MTL) revealed by recent seismic reflection studies, southwest Japan. Tectonophysics, 264, 51—63. ITOH, Y., TSUTSUMI, H., YAMAMOTO, H. & ARATO, H. 2002. Active right-lateral strike-slip fault zone along the southern margin of the Japan Sea. Tectonophysics, 351, 301-314. IWAMURA, K. 2001. Tectonics of the Kake Fault developed in western Hiroshima Prefecture, southwest Japan. Master Thesis, Yamaguchi University. JOLIVET, L., SHIBUYA, H. & FOURNIER, M. 1995. Paleomagnetic rotations and the Japan Sea opening. In: Taylor, B. & Natland, J. (eds) Active Margins and Marginal Basins of the Western Pacific. AGU Geophysical Monograph, 88, 355-369. KAMATA, H. 1998. Quaternary volcanic front at the junction of the Southwest Japan arc and the Ryukyu arc. Journal of Asian Earth Sciences, 16, 67-75. KANAORI, Y. 1990. Late Mesozoic-Cenozoic strikeslip and block rotation in the inner belt of Southwest Japan. Tectonophysics, 111, 381-399. KANAORI, Y. 1999. Reactivation of cataclastic belts and formation of fault gouge in the western

Chugoku region. Gekkan Chikyu (The Earth Monthly}, 21, 22-29. KANAORI, Y., ENDO, Y., YAIRI, K. & KAWAKAMI, S. 1990. A nested fault system with block rotation caused by left-lateral faulting: the Neodani and Atera faults, central Japan. Tectonophysics, 111, 401-418. KANAORI, Y., KAWAKAMI, S. & YAIRI, K. 1992. The block structure and Quaternary strike-slip block rotation of central Japan. Tectonics, 11, 47-56. KANAORI, Y., KOBAYASHI, K., ANNO, Y., WARIGAYA, T. & YAMAMOTO, T. 1999. House damage and seismogenic fault of the 1997 magnitude 6.1 northern Yamaguchi Prefecture earthquake. Journal of the Japan Society of Engineering Geology, 40, 220-230. KlMINAMI, K., MlYASHITA, S. & KAWABATA, K. 1994.

Ridge collision and in situ greenstones in accretionary complexes: an example from the Late Cretaceous Ryukyu Islands and southwest Japan margin. The Island Arc, 3, 103-111. KIMURA, G. 1997. Cretaceous episodic growth of the Japanese Islands. The Island Arc, 6, 52-68. KIMURA, G. & MUKAI, A. 1991. Underplated units in an accretionary complex: the Shimanto Belt of Eastern Shikoku, Southwest Japan. Tectonics, 10, 31-50. KINOSHITA, O. 1995. Migration of igneous activities related to ridge subduction in Southwest Japan and East Asian continental margin from the Mesozoic to the Paleogene. Tectonophysics, 245, 25-35. KINOSHITA, O. & ITO, H. 1986. Migration of Cretaceous igneous activity in Southwest Japan related to ridge subduction. Journal of the Geological Society of Japan, 92, 723-735. KUHN, D. & REUTHER, C.D. 1999. Strike-slip faulting and nested block rotations: structural evidence from the Cordillera de Domeyko, northern Chile. Tectonophysics, 313, 383-398. LEMISZKI, P.J. & BROWN, L.D. 1988. Variable crustal structure of strike-slip fault zones as observed on deep seismic reflection profiles. Geological Society of America Bulletin, 100, 665—676. LIN, A. & UDA, S. 1996. Morphologic characteristics of the earthquake surface ruptures on Awaji Island associated with the 1995 southern Hyogo Prefecture earthquake. The Island Arc, 5, 1-15. LIN, A., IMIYA, H., et al. 1995. Investigation of the Nojima earthquake fault that occurred on Awaji island in the southern Hyogo Prefecture earthquake. Journal of Geography, 104, 113-126. LUYENDYK, B.P., KAMERLING, M.J. & TERRES, R.R. 1980. Geometric model for Neogene crustal rotations in southern California. Geological Society of America Bulletin, 91, 211-217. MACKENZIE, J.S., NEEDHAM, D.T. & AGAR, S.M. 1987. Progressive deformation in an accretionary complex: an example from the Shimanto belt of eastern Kyushu, southwest Japan. Geology, 15, 353-356. MANDL, G. 1987. Tectonic deformation by rotating parallel faults - the 'bookshelf mechanism. Tectonophysics, 141, 277-316.

DISTRIBUTED STRIKE-SLIP FAULTING IN JAPAN MARTEL, S.J., POLLARD, D.D. & SEGALL, P. 1988. Development of simple strike-slip fault zones, Mount Abbot quadrangle, Sierra Nevada, California. Geological Society of America Bulletin, 100, 1451-1465. MARUYAMA, S. & SENO, T. 1986. Orogeny and relative plate motions: examples of the Japanese islands. Tectonophysics, 127, 305-329. MARUYAMA, S., BANNO, S., MATSUDA, T. & NAKAJIMA, T. 1984. Kurosegawa zone and its bearing on the development of the Japanese islands. Tectonophysics, 110, 47-60. MARUYAMA, S., ISOZAKI, L, KIMURA, G. & TERABAYASHI, M. 1997. Paleogeographic maps of the Japanese Islands: plate tectonics synthesis from 750 Ma to the present. The Island Arc, 6, 121-142. MATSUNAGA, S. 1997. Geometry and kinematics of the strike-slip fault system around Yamaguchi City, western Chugoku region. Graduation Thesis, Yamaguchi University. MATSUOKA, A. 1992. Jurassic-Early Cretaceous evolution of the Southern Chichibu terrane, southwest Japan. Palaeogeography, Palaeoclimatology, Palaeoecology, 96, 17-88. MATSUOKA, A. & YAO, A. 1990. Southern Chichibu terrane. In: ICHIKAWA, K., MIZUTANI, S., HARA, L, HADA, S. & YAO, A. (eds) Pre-Cretaceous Terranes of Japan. Publication of IGCP Project 224, 203-216. McKENZiE, D.P. & JACKSON, J. 1983. The relationship between strain rates, crustal thickening, palaeomagnetism, finite strain and fault movements within a deforming zone. Earth and Planetary Science Letters, 65, 182-202. MiYATA, T. 1990. Slump strain indicative of paleoslope in Cretaceous Izumi sedimentary basin along Median Tectonic Line, Southwest Japan. Geology, 18, 392-394. MIZUTANI, S. 1990. Mino terrane. In: ICHIKAWA, K., MIZUTANI, S., HARA, L, HADA, S. & YAO, A. (eds) Pre-Cretaceous Terranes of Japan. Publication of IGCP Project 224, 121-135. MOORE, M.J. 1979. Tectonics of the Najd transcurrent fault system, Saudi Arabia. Journal of the Geological Society, London, 136, 441-454. MURAKAMI, N. 1975. Late Mesozoic to Paleogene igneous activity in West Chugoku, Southwest Japan. Journal of the Geological Society of Japan, 91, 723-742. MURAKAMI, N., IMAOKA, T. & GANZAWA, Y. 1989. Some problems concerning the distribution and geological age of Paleogene cauldrons in West San-in, Southwest Japan. Chidanken Senpou (Special Report of the Association for Geological Collaboration in Japan), 36, 41-47. MURATA, A. 1982. Large Decke structures and their formative process in the Sanbagawa-Chichibu, Kurosegawa and Sambosan terrains, southwest Japan. Journal of the Faculty of Science, University of Tokyo (section If), 20, 277-293. MURATA, A. 1990. Displacement distribution along the Kaminirogawa Fault and a dislocation model. Journal of Geography, 99, 64-75.

163

NAKAJIMA, T. 1997. Regional metamorphic belts of the Japanese islands. The Island Arc, 6, 69-90. NAKAJIMA, T., ISHIWATARI, A., SANO, S. et al. 1992. Geotraverse across the Southwest Japan arc: An overview of tectonic setting of Southwest Japan. In: Metamorphic Belts and Related Plutonism in the Japanese Islands, IGC '92 Kyoto, Field Trip Guidebook, 5, 171-253. Geological Survey of Japan. NAKAJIMA, T., SHIRAHASE, T. & SHIBATA, K. 1990. Along-arc lateral variation of Rb-Sr and K-Ar ages of Cretaceous granitic rocks in Southwest Japan. Contributions to Mineralogy and Petrology, 104,381-389. NATAL'IN, B.A., PARFENOV, L.M., VRUBLEVSKY, A.A., KARSAKOV, L.P. & YUSHMANOV, V.V. 1986. Main fault systems of the Soviet Far East. Philosophical Transactions, Royal Society of London, 317, 267-275. NICHOLSON, C, SEEBER, L., WILLIAMS, P. & SYKES, L. 1986. Seismic evidence for conjugate slip and block rotation within the San Andreas fault system, southern California. Journal of Geophysical Research, 91, 4891-4908. NIITSUMA, N. 1989. Collision tectonics in the southern Fossa Magna, central Japan. Modern Geology, 14, 3-18. NISHIMURA, Y. 1990. 'Sangun metamorphic rocks': terrane problem. In: ICHIKAWA, K., MIZUTANI, S., HARA, L, HADA, S. & YAO, A. (eds) Pre-Cretaceous Terranes of Japan. Publication of IGCP Project 224, 63-79. NISHIMURA, Y. 1998. Geotectonic subdivision and areal extent of the Sangun belt, Inner Zone of Southwest Japan. Journal of Metamorphic Geology, 16, 129-140. NISHIMURA, Y. & IMAOKA, T. 1995. Compilation of radiometric ages of rocks from the Yamaguchi Prefecture. Journal of the Geological Association of Yamaguchi, 34, 1-13. NISHIMURA, Y., IMAOKA, T., UTAMURA, S. & KAMETANI, T. 1995. New Edition of the Geological Map of Yamaguchi Prefecture (1/150000). Geological Association of Yamaguchi. OHMI, S. 2002. Aftershock distribution of the 2002 Western Tottori earthquake by use of the station corrections as a function of hypocenter coordinates. Journal of the Seismological Society of Japan, 54, 575-580. OHTOMO, Y. 1993. Origin of the Median Tectonic Line. Journal of Science of the Hiroshima University, 9, 611 -669. OKADA, A. 1980. Quaternary faulting along the Median Tectonic Line of Southwest Japan. In: Ichikawa, K. (ed.) Median Tectonic Line of Southwest Japan. Memoirs of the Geological Society of Japan, 18, 79-108. OKADA, A. 1992. Proposal of the segmentation of the Median Tectonic Line active fault system. Memoirs of the Geological Society of Japan, 40, 15-30. OTOFUJI, Y. & MATSUDA, T. 1983. Paleomagnetic evidence for the clockwise rotation of Southwest

164

O. FABBRI£TAL.

Japan. Earth and Planetary Science Letters, 62, 349-359. OTOFUJI, Y. & MATSUDA, T. 1984. Timing of rotational motion of Southwest Japan inferred from paleomagnetism. Earth and Planetary Science Letters, 70, 373-382. OTOFUJI, Y. & MATSUDA, T. 1987. Amount of clockwise rotation of Southwest Japan - fan shape opening of the southwestern part of the Japan Sea. Earth and Planetary Science Letters, 85, 289-301. OTOFUJI, Y., MATSUDA, T. & NOHDA, S. 1985. Opening mode of the Japan Sea inferred from the palaeomagnetism of the Japan Arc. Nature, 317, 603-604. OTOFUJI, Y., ITAYA, T. & MATSUDA, T. 1991. Rapid rotation of southwest Japan - palaeomagnetism and K-Ar ages of Miocene volcanic rocks of southwest Japan. Geophysical Journal International, 105, 397-405. OTSUKI, K. 1992. Oblique subduction, collision of microcontinents and subduction of oceanic ridges: their implications on the Cretaceous tectonics of Japan. The Island Arc, 1, 51-63. RANDALL, D.E., TOMLINSON, AJ. & TAYLOR, O.K. 2001. Paleomagnetically defined rotations from the Precordillera of northern Chile: evidence of localized in situ fault-controlled rotations. Tectonics, 20, 235-254. RESEARCH GROUP FOR ACTIVE FAULTS OF JAPAN 1980. Active Faults in Japan. Sheet maps and inventories. Tokyo University Press. RESEARCH GROUP FOR ACTIVE FAULTS OF JAPAN 1991. Active Faults in Japan. Sheet maps and inventories. Revised Edition, Tokyo University Press. RON, H., FREUND, R. & GARFUNKEL, Z. 1984. Block rotation by strike-slip faulting: structural and paleomagnetic evidence. Journal of Geophysical Research, 89, 6256-6270. SAGIYA, T., NISHIMURA, T., HATANAKA, Y., FUKUYAMA, E. & ELLSWORTH, W.L. 2002. Crustal movements associated with the 2000 western Tottori earthquake and its fault model. Journal of the Seismological Society of Japan, 54, 523-534. SAKAKIBARA, N. 1995. Structural evolution of multiple ductile shear zones system in the Ryoke belt, Kinki Province. Journal of Science of the Hiroshima University, 10, 267-332. SCHREURS, G. 1994. Experiments on strike-slip faulting and block rotation. Geology, 22, 567-570. SENO, T., STEIN, S. & GRIPP, A.E. 1993. A model for the motion of the Philippine Sea plate consistent with NUVEL-1 and geologic data. Journal of Geophysical Research, 102, 53-84. SHIBATA, K. & TAKAGI, H. 1988. Isotopic ages of rocks and intrafault materials along the Median Tectonic Line - an example in the Bungui-toge area, Nagano Prefecture. Journal of the Geological Society of Japan, 94, 35-50. STEIN, G., CHARVET, J., LAPIERRE, H. & FABBRI, O. 1994. Geodynamic setting of volcano-plutonic rocks in so-called 'paleo-accretionary prisms':

forearc activity or post-collisional magmatism? Lithos,33, 85-101. TAGAMI, T., LAL, N. & NISHIMURA, S. 1988. Fission track thermochronologic analysis of the Ryoke belt and the Median Tectonic Line, southwest Japan. Journal of Geophysical Research, 95, 13705-13715. TAIRA, A. & OGAWA, Y. 1988. The Shimanto Belt, Southwest Japan: studies on the evolution of an accretionary prism. Modern Geology, 12, 1-542. TAIRA, A. TASHIRO, M., OKAMURA, M. & KATTO, J. 1980. The geology of the Shimanto Belt in Kochi Prefecture, Shikoku, Japan. In: TAIRA, A. & TASHIRO, M. (eds) Geology and Paleontology of the Shimanto Belt. Rinya-Kosaikai Press, Kochi, 319-389. TAIRA, A., OKADA, H., WHITAKER, J.H.M. & SMITH, AJ. 1982. The Shimanto belt of Japan: Cretaceous-Lower Miocene sediments in forearc basin to deep trench environments. In: LEGGETT, J. (ed.) Trench Forearc Geology: Sedimentation and Tectonism on Modern and Active Plate Margins. Geological Society, London, Special Publication, 10, 5-26. TAIRA, A., SAITO, Y. & HASHIMOTO, M. 1983. The role of oblique subduction and strike-slip tectonics in the evolution of Japan. In: HILDE, T.W.C. & UYEDA, S. (eds) Geodynamics of the Western Pacific-Indonesian Region. AGU Geodynamic series, 11, 303-316. TAKAGI, H. 1986. Implications of mylonite microstructures for the geotectonic evolution of the Median Tectonic Line, central Japan. Journal of Structural Geology, 8, 3—14. TAKAGI, H. & SHIBATA, K. 1992. K-Ar dating of fault gouge - examples along the Median Tectonic Line active fault system. Memoirs of the Geological Society of Japan, 40, 31-38. TAKASU, A., WALLIS, S.R., BANNO, S. & DALLMEYER, R.D. 1994. Evolution of the Sanbagawa metamorphic belt, Japan. Lithos, 33, 119-133. TAMAKI, K., SUYEHIRO, K., ALLAN, J., INGLE, J.C., & PISCIOTTO, K. 1992. Tectonic synthesis and implications of Japan Sea OOP drilling. Proceedings Ocean Drilling Program, Scientific Results, 127-128, 1333-1350. TAYLOR, G.K., GROCOTT, J., POPE, A. & RANDALL, D.E. 1998. Mesozoic fault systems, deformation and fault block rotation in the Andean forearc: a crustal scale strike-slip duplex in the Coastal Cordillera of northern Chile. Tectonophysics, 299, 93-109. TIKOFF, B., TEYSSIER, C. & WATERS, C. 2002. Clutch tectonics and the partial attachment of lithospheric layers. EGS Stephan Mueller Special Publication Series, 1, 93-117. TSUKAHARA, H. & KOBAYASHI, Y. 1991. Crustal stress field in central and southwest Japan. Journal of the Seismological Society of Japan, 44,221-231. TSUTSUMI, H., OKADA, A., NAKATA, T., ANDO, M. & TSUKUDA, T. 1991. Timing and displacement of Holocene faulting on the Median Tectonic Line in central Shikoku, southwest Japan. Journal of Structural Geology, 13, 227-233.

DISTRIBUTED STRIKE-SLIP FAULTING IN JAPAN UYEDA, S. & MIYASHIRO, A. 1974. Plate tectonics and the Japanese islands: a synthesis. Geological Society of America Bulletin, 85, 1159-1170. WALLIS, S. & BANNO, S. 1990. The Sanbagawa belt trends in research. Journal of Metamorphic Geology, 8, 393-399. WIBBERLEY, C.AJ. & SfflMAMOTO, T. 2003. Internal structure and permeability of major strike-slip fault zones: the Median Tectonic Line in Mie Prefecture, Southwest Japan. Journal of Structural Geology, 25, 59-78. YAMADA, N., HIGASHIMOTO, S., MIZUNO, K., HIROSHIMA, T. & SUDA, Y. 1986.1/200 000 Geological Map of the Hiroshima District. Geological Survey of Japan. YAMAKITA, S. & ITO, T. 1999. A numerical model of the subsidence process of the Izumi sedimentary basin. Structural Geology (The Journal of the Tectonic Research Group of Japan), 43, 1-9. YAMAKITA, S. & OTOH, S. 2000. Cretaceous rearrangement processes of pre-Cretaceous geologic units of the Japanese Islands by MTL-Kurosegawa left-lateral strike-slip fault system. Memoirs of the Geological Society of Japan, 56, 23—28. YAMAKITA, S., ITO, T., TANAKA, H. & WATANABE, H. 1995. Early Oligocene top-to-the-west motion along the Sashu fault, a low-angle oblique thrust of the paleo-Median Tectonic Line, east Kyushu,

165

Japan. Journal of the Geological Society of Japan, 101, 978-988. YAMAMOTO, H. 1993. Submarine geology and postopening movements in the southern region of the Sea of Japan. Marine Geology, 112, 133-150. YAMAMOTO, H. 1994. Kinematics of mylonitic rocks along the Median Tectonic Line, Akaishi range, central Japan. Journal of Structural Geology, 16, 61-70. YOON, S.H. & CHOUGH, S.K. 1995. Regional strikeslip in the eastern continental margin of Korea and its tectonic implications for the evolution of Ulleung Basin, East Sea (Sea of Japan). Geological Society of America Bulletin, 107, 83-97. YOSHIKAWA, S., IWASAKI, Y., IKAWA, T., KURODA, T. &

YOKOTA, Y. 1992. Geological structure of the MTL in western Wakayama by seismic reflection study. Memoirs of the Geological Society of Japan, 40, 177-186. YOSHIKURA, S., HADA, S. & ISOZAKI, Y. 1990. Kurosegawa terrane. In: ICHIKAWA, K., MIZUTANI, S., HARA, I., HADA, S. & YAO, A. (eds) Pre-Cretaceous Terranes of Japan. Publication of IGCP Project 224, 185-201. YUSA, Y., TAKEMURA, K., etal 1992. Subsurface structure of Beppu Bay (Kyushu, Japan). Journal of the Seismological Society of Japan, 45, 199-212.

This page intentionally left blank

Growth and demise of continental arcs and orogenic plateaux in the North American Cordillera: from Baja to British Columbia DONNA L. WHITNEY1, SCOTT R. PATERSON2, KEEGAN L. SCHMIDT2'4, ALLEN F. GLAZNER3 & CHRISTOPHER F. KOPF3 1 Geology and Geophysics, University of Minnesota, Minneapolis, Minnesota 55455, USA (e-mail: [email protected]) 2 Earth Sciences, USC, Los Angeles, California 90089, USA ^Geological Sciences, UNC-Chapel Hill, North Carolina 27599, USA 4 Present address: Natural Sciences, Lewis-Clark State College, Lewiston, Idaho 83501, USA Abstract: In the North American Cordillera, crustal thickening, magmatism and flow of deep crust created an orogenic plateau, or series of related plateaux, in the Late Mesozoic-Early Cenozoic. From west to east, the plateaux extended from the continental arcs to the inboard crystalline belts of the Omineca-Sevier belt. From north to south, the plateaux ranged from British Columbia/SE Alaska to Baja California, Mexico. Although a vast region of western North America was characterized by thickened crust (60-70 km), unroofing of deep crust from > 30 km was largely confined to the edges of the plateaux: the arcs and the eastern margins. Comparison of the unroofing histories of the Cordilleran arcs reveals that they differed dramatically from each other in the amount and style, but not timing, of exhumation. The northern Cordilleran arc and northern interior (Omineca) belt were exhumed from deep mid-crustal levels, with regional-scale Eocene extension accompanied by magmatism. In contrast, the central (Sierra Nevada) and southern (Peninsular Ranges) arcs were unroofed to much shallower levels (typically <15 km), primarily by erosion and local deformation. North to south differences in exhumation style and magnitude in the Cordilleran arcs may reflect differences in the degree of coupling between the subducting plate and the thickened continental lithosphere in the north v. south. In the northern Cordillera, relationships between Pacific-region plate activity and Tertiary continental extension/magmatism and deep exhumation suggest continued geodynamic coupling between subducting plates and orogenic crust following crustal thickening and plateau formation. In contrast, the central and southern Cordilleran arcs do not contain evidence for mechanical links with the subducting plate after the Late Cretaceous.

The subduction of oceanic lithosphere beneath continents has a profound effect on continental lithosphere by driving crustal thickening and magmatism, and is a precursor of collision tectonics. Of interest is how deformation of the continental lithosphere and the mechanical relationship of the continent to the subducting plate evolve through time following maximum crustal thickening and subduction-related magmatism. In particular, what is the geodynamic relationship between plate tectonic factors (e.g. velocities and trajectories of oceanic plates) and processes internal to continental orogenic crust (e.g. gravitational instability of heated, thickened crust)? Previous studies have proposed varying degrees of coupling between subducting plates and thickened continental lithosphere, including recent work emphasizing a large degree of decoupling (e.g. Vanderhaeghe & Teyssier 2001).

The origin and fate of orogenic plateaux may be related to these issues of coupling/decoupling of subducting plates and thickened continental lithosphere. Orogenic plateaux are a first-order expression of convergent plate boundaries involving continental lithosphere (e.g. present-day Tibet and the Andean Altiplano-Puna). Plateau construction may involve lithosphere-scale shortening of both crust and upper mantle, or may involve decoupling of the crust from the mantle, with shortening taken up by the rheologically weak deep crust. The mechanisms and rates of construction and demise of orogenic plateaux provide information about the thermo-mechanical behaviour of continental lithosphere from the mantle to the surface - during and after plate convergence. In this chapter, we examine the exhumed margins of a palaeoplateau for evidence of the degree of coupling

From: GROCOTT, J., MCCAFFREY, K. J. W., TAYLOR, G. & TIKOFF, B. (eds) 2004. Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 167-175. 0305-8719/04/$15 © The Geological Society of London 2004.

168

D. L. WHITNEY ET AL

between the subducting oceanic plate and the thickened continental lithosphere. A large plateau likely existed in the interior (Sevier belt, Fig. 1) of the North American Cordillera in the Late Cretaceous-Early Tertiary, analogous to orogenic plateaux in active orogens,

Fig. 1. Generalized map of the North American Cordillera showing the location of the batholith belts (grey shading) and the inboard orogenic zone (Omineca-Sevier belt). C, Cascades. Black shading shows location of metamorphic core complexes. Stippled region shows location of former thickened crust (>60 km). Stars show general location of rocks recording >9 kbar (arc/batholith belts only; data not shown for Omineca-Sevier belt). The western margin of the plateau was approximately at the western margin of the arcs, but was located within the PRB at the boundary between western and eastern zones (see Fig. 5).

as has been proposed by many authors (e.g. Coney & Harms 1984; England & Thompson 1986; Molnar & Lyon-Caen 1988; Wernicke et al 1996; Wolfe et al 1998; Dilek & Moores 1999). In this chapter, we extend this idea and propose that the plateau regions included the arcs in the northern, central and southern Cordillera, as well as the inboard Sevier and Omineca belts (Figs 1 & 2a). To understand the evolution of these plateau regions, we investigate the record of crustal thickening and unroofing at the western margin of the plateaux: the continental arcs. Cordilleran crustal thickening and unroofing Large tracts of the western continental margin from Baja, Mexico, to British Columbia and SE Alaska are dominated by rocks formed during Mesozoic-Early Cenozoic magmatism, metamorphism and deformation. The continental arcs, now unroofed batholith belts, include the Coast Mountains-Cascades (CMC), the Sierra Nevada (SN) and the Peninsular Ranges batholith (PRB). A parallel belt of crystalline rocks occurs inboard of the arcs (Omineca belt in the north; Sevier belt in the south) (Fig. 1). The batholith belts developed as Andean-style arcs generated by east-dipping subduction. The exhumed continental arcs in the North American Cordillera are broadly similar in that they were generated in part by east-dipping, Mesozoic-Early Cenozoic subduction of oceanic plates (Table 1). The timing, style (erosion v. tectonic denudation) and magnitude of unroofing/exhumation of the different batholith belts, however, varied from north to south, with the most dramatic evidence for deep exhumation of arc metamorphic rocks and high-pressure plutonic rocks in the northern Cordillera (Whitney et al 1999; Valley et al 2003) (Table 1, Fig. 2). It is important to note that the exposure of formerly deep rocks at the surface may indicate the former presence of thickened crust, but the absence of these rocks at the surface does not necessarily indicate thinner crust. Although the Tertiary tectonic evolution of the Sevier-Omineca belt, inboard of the arcs, also varied from north to south, we focus in this chapter on the continental arcs because they occupied a key tectonic position for evaluating dynamic links between the subducting plates and the continents. In particular, if the arcs shared similar constructional histories, the timing and mechanisms of unroofing of the arcs can be compared and used to evaluate the relative roles of external v. internal processes in the evol-

CORDILLERAN ARC AND OROGENIC PLATEAUX

169

exhumed mid-crust (6-7 kbar; Friedman & Armstrong 1988). Evidence for thickened crust in the Late Cretaceous includes: (1) the presence of 12 kbar (40 km), upper amphibolite facies rocks exhumed at the surface, indicating that the crust must once have been thicker than at present (32-35 km; Potter et al 1986; Cook et al 1992; Clowes et al 1995); and (2) the geochemistry of calc-alkaline plutons indicates genNorthern Cordillera eration by partial melting of garnet-bearing mafic The Coast Mountains-Cascades (CMC) orogen crust at depths of ^60 km (DeBari et al 1998). is a > 1500 km long belt that extends from The active magmatic arc therefore contained Washington to SE Alaska (Fig. 1). The CMC thickened crust (60-70 km) in the Late Cretacformed during mid-Cretaceous accretion of ter- eous (Miller & Paterson 2001). In the CMC arc, exhumation commenced ranes and east-dipping subduction beneath the accreted terranes. Intrusive activity occurred epi- during Late Cretaceous contraction (Miller & sodically from 110 to 45 Ma, including a major Paterson 2001), and both the CMC and the magmatic episode in the Tertiary that produced Omineca belts were thinned by Eocene extenbatholith-sized intrusions in the arcs (Paterson sion. The Omineca belt is located between the & Miller 1998, and references therein; Stowell Intermontane belt and the palaeo-margin of & McClelland 2000, and references therein) North America (Fig. 1), and consists of a series of metamorphic core complexes that formed (Table 1) and inboard Omineca belt (Fig. 1). The Cascade Range, at the southern end of the during Eocene extension (Parrish et al 1988). Coast Mountains, is offset from the rest of the The core complexes contain migmatitic domes belt by strike-slip faults (Fig. 1). Differences in that record high-T metamorphism (up to the large-scale structure and exhumation history ~800°C) at mid-crustal pressures (7-10 kbar) of the Cascades compared to the rest of the (e.g. Spear & Parrish 1996; Norlander et al Coast belt may be due to the position of the Cas- 2002). Based on metamorphic data and geocades at the 'end' of the orogen. For example, the dynamic considerations (discussed below), we overall antiformal structure of the Cascades may propose that the entire region from the southern represent buckling at an orogenic corner (syn- CMC to the Omineca belt was thickened to at taxis), analogous to the western and eastern least 60 km, primarily in the Late Cretaceous. ends of the Himalayan orogen. Despite some structural differences, the Cascades and Coast Mountains both contain moder- Central Cordillera ately high-pressure rocks (Figs 1 & 3), including Magmatism in the ~600 km long Sierra Nevada meta-supracrustal rocks buried to depths of 30- belt occurred between ^220 and 80 Ma, with the 40 km. Some regions along the margins of the greatest volume between 98-86 Ma (Bateman crystalline core record low-P metamorphism, but 1992; Coleman & Glazner 1997) (Table 1). moderately high-pressure rocks (~9-12 kbar) Emplacement depths of plutons decrease from have been reported in SE Alaska, northern and 9-15 km in the west to <3km in the east southern British Columbia, and throughout the (Ague & Brimhall 1988) (Fig. 4). Recorded Cascades (Whitney et al 1999, and references pressures of metamorphic rocks are typically therein) (Figs 1-3). Local regions in the Inter- low (1.5-3 kbar), but are higher (~6-9 kbar) montane belt to the east of the arc also comprise at the southern end of the batholith where ution and eventual demise of orogenic plateaux. In the following sections, we summarize the history and characteristics of each region of the North American Cordillera (northern, central and southern) to determine the magnitude of crustal thickening, and timing and mechanism(s) of unroofing.

Table 1. Comparison of the Coast Mountains-Cascades (CMC), Sierra Nevada (SN) and Peninsular Ranges batholith (PRB) belts* Arc CMC SN PRB

Timing of major magnetism (Ma)

Max. P (kbar) of exposed rocks

Max. crustal thickness1^ (km)

Modern crustal thickness (km)

110-45 98-86 164-85

12 9 (but typically <5) <6

>60 40-45 or 60-70 >55

32-35 33 25-43

*See text for references. t Inferred (see text).

170

D. L.WHITNEY ET AL

Fig. 2. (a) Cartoon showing the spatial relationships of the arc, interior belt, thickened crust and the plateau, with the modern and Late Cretaceous moho depths shown, and the characteristic exposure levels of the Coast Mountains-Cascades (CMC), Sierra Nevada (SN) and Peninsular Ranges batholith (PRB). (b) Pressuretemperature diagram showing representative data for each of the three arcs discussed in this chapter. See text for references.

additional unroofing has occurred due to faulting (Pickett & Saleeby 1993) (Figs 2 & 4). These variations reflect differential unroofing, rather than variations in Late Cretaceous crustal thickness. The western margin of crustal thickening in the Sierra Nevada was the Western Metamorphic Belt (WMB; Fig. 4), which was thickened between 155 and 112 Ma (Ague & Brimhall 1988; Paterson et al. 1991). West of this belt,

marine sediments were deposited from Late Jurassic through Tertiary time, but the region to the east comprised thickened crust (Wernicke et al 1996; Wolfe et al 1998). The presence of Cretaceous volcanic sequences throughout the eastern Sierra and the emplacement depths of plutons indicate that 510 km of crust has been removed since the Cretaceous. Ague & Brimhall (1988) estimated that 6 km of Sierran crust had been removed by erosion. These data and estimates indicate a palaeo-crustal thickness of ~40-45 km, but Wernicke et al (1996) noted that the presentday crust, from the eastern Sierra to the Basin and Range, has been thinned by 50% over the last 20 Ma. The present-day Moho is ~33 km under the Sierra (Wernicke et al 1996). This implies a crustal thickness of 60-70 km prior to extension, similar to the crustal thickness estimates derived from xenolith geobarometry (Ducea & Saleeby 1996). Southern Cordillera The Peninsular Ranges batholith (PRB) extends for 1600km from southern California to the southern tip of Baja California, Mexico (Figs 1 & 5). It developed as a magmatic arc in the Jurassic and Cretaceous (~ 164-85 Ma) along a Neoproterozoic continental margin (DePaolo 1981; Silver & Chappell 1988; Gastil et al 1990; Schmidt & Paterson 2002). Cenozoic cover obscures most of the PRB south of 28° latitude, but the northern 700 km are well exposed. The batholith is underlain by distinct western (oceanic) and eastern (continental) belts, as

Fig. 3. Maps of (a) the northern and (b) the southern segments of the Coast Mountains-Cascades belt showing the locations of metamorphic and plutonic rocks with recorded pressures of >9 kbar (stars), (a) is modified from Stowell & Crawford (2000).

CORDILLERAN ARC AND OROGENIC PLATEAUX

Fig. 4. Map of the Sierra Nevada batholith showing the emplacement depths of plutons and estimated metamorphie pressures for country rocks and roof pendants. Map modified from Bateman (1992), with data from Ague & Brimhall (1988), Hanson et al (1993) and Pickett & Saleeby (1993). The thick black dashed line at the western margin of the WMB marks the western limit of crustal thickening. WMB, Western Metamorphie Belt.

identified by the petrology and geochemistry of plutons (Silver & Chappell 1988) and prebatholithic stratigraphy (Gastil 1993). Deformation was focused along the lithospheric boundary between the belts during intrusion of much of the batholith (Johnson et al 1999; Schmidt & Paterson 2002). Recorded metamorphie conditions of exhumed rocks change dramatically across this zone from sub- to lower greenschist facies and P < 2.5 kbar in the west to amphibolite facies and P ~ 4-6 kbar across a broad region in the east (e.g. Todd et al. 1988; unpublished data, this study) (Figs 2 & 5). Geophysical studies suggest contrasting basement in these belts, with faster seismic velocities (Magistrale & Sanders 1995) and a relatively flat moho at 37-41 km beneath the western PRB, and the thickest crust (up to 43 km) beneath the transition zone. To the east, slower velocities and eastward shallowing to ~25 km are related to present-day rifting (Ichinose et al. 1996; Lewis

171

et al 2000). Contrasts in crustal density and thickness between western and eastern zones of the batholith in mid-Cretaceous time resulted in dramatic differences in the amount of exhumation experienced by these two crustal belts. Following >30Ma of structural thickening and arc magmatism, a > 100 km wide region of the eastern PRB was deeply denuded in the Late Cretaceous. Approximately 20km of material was stripped from the eastern zone relative to the western zone during this time, with denudation on the order of 1 mm a"1. Erosion, coupled with local thrust faulting (Schmidt 2000), was by far the most important denudation mechanism; only one Late Cretaceous extensional fault has been found (Erskine & Wenk 1985). Kimbrough et al (2001) proposed that the development of erosional topography in the Late Cretaceous was related to the emplacement of voluminous magmatic bodies. Pre-exhumation reconstruction of the eastern zone at ^100 Ma indicates crustal thickness of the order of ~55 km (Schmidt 2000). This estimate is consistent with petrological evidence for anomalously thick continental crust during production of the volumetrically significant ~99-92 Ma La Posta magmatic suite that intruded a broad region of the eastern and transitional zones of the batholith (Kimbrough et al 2001). An estimate of at least 55 km is also consistent with the modern crustal thickness for the eastern/transition zone (35-40 km) plus the amount estimated to have been removed by erosion (~20 km). Based on these data and inferences, we propose that the eastern PRB formed the western margin of thickened crust - and perhaps an erogenic plateau — in the Late Cretaceous, separated from the western zone by a sharp topographic break. The fore-arc basins along the west coast received voluminous but locally derived material from the eastern PRB during Late Cretaceous-Paleocene time. Not until the Eocene did regional fluvial systems cut through the ancestral Peninsular Ranges and deposit exotic, Sonoran-derived sediment in these basins (Axen et al 2000). The North American Cordilleran plateau(x) Evidence for thickened crust does not directly translate into evidence for high-elevation/lowrelief landscapes. Indications of higher preMid-Eocene elevations in the North American Cordillera come from palaeoaltimetry studies and inferences about crustal dynamics. In the northern Cordillera, Wolfe et al (1998) used

172

D. L.WHITNEY ET AL.

Fig. 5. (a) Map of the Peninsular Ranges batholith, with (b) a more detailed map of part of the belt. ABF, Agua Blanca fault; SSPM, Sierra San Pedro Martir. The thick dashed line marks the transition from magnetite-bearing (Mt) plutons to ilmenite-bearing (Ilm) plutons. The black line (solid south of the ABF) is the boundary between low-P rocks with > 100 Ma cooling ages and moderate-P rocks with <90 Ma cooling ages. This line therefore shows the western edge of the proposed southern Cordillera plateau.

palaeoaltimetry based on fossil plant assemblages to propose that northeastern Washington and southern British Columbia were at least 1-2 km higher in the early Mid-Eocene (^50 Ma) than today. In the central Cordillera, palaeobotanical evidence also suggests that parts of the western USA were at least 2 km higher in the Mid-Eocene than they are today (Wolfe et al. 1998). Palaeoaltimetry data, combined with tectonic evidence for magnitude of crustal thickness and subsequent thinning (Wernicke et al. 1996; Ducea & Saleeby 1996), indicate that the Sierra may have been at elevations of 4-5 km in the Late Cretaceous. The present-day average height of the Sierran crest is 2.8 km. Although there are no comparable data for palaeo-elevations in Baja, Mexico, isostatic models for the Peninsular Ranges suggest that crust of the thickness (>55 km) and density of the eastern zone could have supported average surface elevations of >4km over a broad region (Schmidt 2000). In contrast, the dense, relatively little denuded, and presently anomalously thick western zone probably could not have supported more than 1 km of average surface elevation during the Cretaceous. We therefore infer that the eastern PRB represented a zone of high elevation in the Late Cretaceous. Comparison with the modern Andean continental arc is useful for assessing the relationship between crustal thickening and topography in the North American Cordilleran arcs. In the Andes, the ~4km high, 350-400 km wide Altiplano and Puna Plateaux were created by thickening of lithosphere during subduction.

Crustal thickness varies, but is generally 5070 km (Zandt et al 1994). The plateau region is bounded on the eastern side by a thrust belt, and on the west by the arc. The Altiplano-Puna Plateau comprises varied and complex basement geology, as does the North American Cordillera, and has been uplifted across a wide area due to the thermo-mechanical effects of contraction of hot, thermally weakened lithosphere. Regional uplift of the plateau regions has been attributed to an increase in plate convergence rate, which led to a decrease in the angle of subduction, and thinning/weakening of the South American lithosphere (Allmendinger et al 1997). In the Cordillera, the CMC and Sierran belts likely comprised the western margins of broad plateau regions, similar to the relation between the Andean arc and the Altiplano-Puna Plateau. The PRB may have represented the southern end of the plateau, analogous to the narrow, southern end of the Altiplano-Puna Plateau. It is difficult to identify the northern end of the plateau because of lack of data and effects of faulting in the northern Cordillera. These observations and their similarities to inferences about the Mesozoic-Early Cenozoic North American arcs support the idea that plateau formation commonly accompanies construction of continental arcs. Unroofing mechanisms The unroofing of the North American Cordilleran arcs and inboard regions can be described in terms of the collapse of the plateau region (east of the

CORDILLERAN ARC AND OROGENIC PLATEAUX

arcs) and the unroofing of the arcs. The former is characterized by metamorphic core complex development and, in the central and southern Cordillera, by Basin and Range style extension. Unroofing of the arcs occurred by a combination of erosion and syn- to post-convergence extension (transpression, then transtension). The balance between these processes varied in the northern v. the central/southern Cordilleran arcs. The Omineca-Sevier belt may have collapsed due to the presence of hot, thickened crust (Livaccari 1991) weakened either by magmatism (Armstrong & Ward 1991) or by partial melting at depth (Vanderhaeghe & Teyssier 2001). The exhumation of deep middle crust (~20-50km depth in a thickened orogen) may also have been influenced by the presence of lowerdensity rocks at depth, due to the emplacement of denser terranes on felsic basement, and/or to the generation of partial melting at depth during heating and decompression (Teyssier & Whitney 2002). These are all internal processes that are consequences of crustal thickening. The arcs also consisted of hot crust of similar thickness to the internal zone, but only the northern belt experienced major collapse and extension. The Sierra and PRB arcs may have been comparable to the CMC in terms of crustal thickening and magmatic history, but they did not experience major extension, and only underwent about half the magnitude of exhumation. This observation is incompatible with a model of gravitational collapse driving unroofing of the arcs, because collapse of hot, thickened crust did not occur on a large scale in the central/southern Cordilleran arcs. This suggests that external mechanisms for driving exhumation may have been important in the north but not in the central and southern regions. External mechanisms include a change in plate motion factors in the Eocene; for example, plate velocity, trajectory and/or number of small plates in the northern Pacific. Gravitational collapse of thickened crust and changes in Pacific plate motions have been proposed to account for extension in the northern Cordillera. For example, rearrangement of Pacific plates to more oblique convergence relative to North America, consumption of the Kula plate and/or a slowing of plate convergence rate in the Eocene (Engebretsen et al 1985; Stock & Molnar 1988) may have caused extension throughout the northern Cordillera. Plate boundary kinematics resulted in a NW-SE minimum principal stress component, consistent with the orientation of syn-extensional structures in the metamorphic terrains of this region. A change in plate boundary stresses (e.g. interaction of the

173

North American margin with the Pacific-Kula— Farallon triple junction) has also long been proposed as a mechanism for Eocene magmatism in the northern Cordillera (Ewing 1980). These observations are consistent with geodynamic coupling between the subducting plate and the continental lithosphere, but the nature of the link cannot be inferred from the observations.

Cordilleran crustal thickening, deep crustal flow and unroofing Previous North American Cordilleran studies (e.g. Coney & Harms 1984; Dilek & Moores 1999) have used exposures of metamorphic rocks to model palaeo-crustal thickness. In these models, regions with exposures of highgrade metamorphic rocks (Omineca-Sevier belt) are depicted as having thicker Late Cretaceous crust than regions with few or no exposures of mid-crustal rocks. For example, these studies show thinner crust (<40 km) in the region between the Omineca and Cascades arc because this belt (Intermontane) generally lacks exposures of high-grade metamorphic rocks. In contrast, we argue that the location of exposed deeper crustal levels is primarily a function of unroofing mechanism(s), and that these mechanisms varied in different regions depending in part on thermo-mechanical interaction between oceanic and continental lithosphere. Thick crust existed in the Late Cretaceous in regions that today do not contain exposures of high-P rocks. This proposal is supported by thermo-mechanical models that predict flow for deep crust on short timescales (McKenzie et al. 2000). Flow from the thickened regions (western arcs and inboard zones) to intervening regions would result in uniformly thick crust and a flat moho, and reduce lateral variations in topography. Lower crustal flow in response to pressure gradients — i.e. away from regions of thickened crust has been proposed to explain the thickening and uplift of the Colorado Plateau (McQuarrie & Chase 2000). Here we propose a similar mechanism for creating a broad region of thickened crust that encompasses much of western North America, from British Columbia to Baja and including the continental arcs, in the Late Cretaceous-Early Tertiary. Such flow implies decoupling of the deep crust from the mantle lithosphere. The subsequent regional-scale response of thickened crust to internal (continental) and external (oceanic plate) factors indicates variable coupling between oceanic and continental plates involved in the construction of the North American Cordillera.

174

D. L. WHITNEY ET AL.

We thank Christian Teyssier, Brendan Murphy and Graeme Taylor for their comments on the paper and the ideas in it. This work was partially supported by NSF grant EAR-9896017 to DLW.

DUCEA, M. & SALEEBY, J. 1996. Buoyancy sources for a large, unrooted mountain range, the Sierra Nevada: evidence from xenolith thermobarometry. Journal of Geophysical Research, 101, 80298044. ENGEBRETSEN, D.C., Cox, A. & GORDON, R.G. 1985. Relative Motions Between Oceanic and ContinenReferences tal Plates in the Pacific Basin. Geological Society of America Special Paper, 206. AGUE, JJ. & BRIMHALL, G.H. 1988. Magmatic arc asymmetry and distribution of anomalous plutonic ENGLAND, P.C. & THOMPSON, A.B. 1986. Some thermal and tectonic models for crustal melting in belts in the batholiths of California; effects of continental collision zones. In: COWARD, M.P. & assimilation, crustal thickness, and depth of crysRIES, A.C. (eds), Collision Tectonics. Geological tallization. Geological Society of America Bulletin, Society, London, Special Publications, 19, 83-94. 100, 912-927. ALLMENDINGER, R.W., JORDAN, T.E., KAY, S.M. & ERSKINE, B.G. & WENK, H.R. 1985. Evidence for Late Cretaceous crustal thinning in the Santa Rosa ISACKS, B.L. 1997. The evolution of the Altiplanomylonite zone, southern California. Geology, 13, Puna plateau of the Central Andes. Annual 274-276. Reviews in Earth and Planetary Sciences, 25, EWING, T.E. 1980. Paleogene tectonic evolution of 139-174. the Pacific northwest. Journal of Geology, 88, ARMSTRONG, R.L. & WARD, P. 1991. Evolving geo619-638. graphic patterns of Cenozoic magmatism in the North American Cordillera: the temporal and FRIEDMAN, R.M. & ARMSTRONG, R.L. 1988. Tatla spatial association of magmatism and metamorphic Lake metamorphic complex: an Eocene metacore complexes. Journal of Geophysical Research, morphic core complex on the southwestern edge 96, 13201-13224. of the Intermontane Belt of British Columbia. AXEN, G.J., GROVE, M. et al 2000. Thermal evolution Tectonics,!, 1141-1166. of Monte Blanco Dome: low-angle normal faulting GASTIL, R.G. 1993. Prebatholithic history of Peninsuduring Gulf of California rifting and late Eocene lar California. In: GASTIL, R.G. & MILLER, R.H. denudation of the eastern Peninsular Ranges. (eds) The Prebatholithic Stratigraphy of PeninsuTectonics, 19, 197-212. lar California. Geological Society of America BATEMAN, P.C. 1992. Plutonism in the Central Part of Special Paper, 279, 145-156. the Sierra Nevada Batholith, California. United GASTIL, R.G., DIAMOND, C.K., WALAWENDER, M.J., States Geological Survey Professional Paper, 1483. MARSHAL, M., BOYLES, C. & CHADWICK, B. CLOWES, R.M., ZELT, C.A., AMOR, J.R. & ELLIS, R.M. 1990. The problem of the magnetite/ilmenite 1995. Lithospheric structure in the southern Canaboundary in southern and Baja California. In: dian Cordillera from a network of seismic refracANDERSON, J.L. (ed.) The Nature and Origin of tion lines. Canadian Journal of Earth Sciences, Cordilleran Magmatism. Geological Society of 32, 1485-1513. America Memoir, 174, 19-32. COLEMAN, D.S. & GLAZNER, A.F. 1997. The Sierra HANSON, R.B., SORENSEN, S.S., BARTON, M.D. & Crest magmatic event; rapid formation of juvenile FISKE, R.S. 1993. Long-term evolution of fluidcrust during the Late Cretaceous in California. rock interactions in magmatic arcs; evidence from International Geology Review, 39, 768-787. the Ritter Range pendant, Sierra Nevada, CaliforCONEY, PJ. & HARMS, T.A. 1984. Cordilleran metania, and numerical modeling. Journal of Petrology, morphic core complexes: Cenozoic extensional 34, 23-62. relics of Mesozoic compression. Geology, 12, ICHINOSE, G., DAY, S.M., MAGISTRALS, H., PRUSH, T., 550-554. VERNON, F.L. & EDELMAN, A. 1996. Crustal thickCOOK, F.A., VARSEK, J.L. et al 1992. Lithoprobe ness variations beneath the Peninsular Ranges, crustal reflection cross section of the southern southern California. Geophysical Research Canadian Cordillera, 1, Foreland thrust and fold Letters, 23, 3095-3098. belt to Fraser River fault. Tectonics, 11, 12-35. DEBARI, S.M., MILLER, R.B. & PATERSON, S.R. 1998. JOHNSON, S.E., TATE, M.C. & FANNING, C.M. 1999. New geologic mapping and SHRIMP U-Pb Genesis of tonalitic plutons in the Cretaceous magzircon data in the Peninsular Ranges batholith, matic arc of the North Cascades: mixing of mantle Baja California, Mexico: evidence for a suture? derived magmas and melting of garnet-bearing Geology, 27, 743-746. lower crust. Geological Society of America KIMBROUGH, D.L., SMITH, D.P. et al. 2001. ForearcAbstracts with Programs, 30, 257-258. basin sedimentary response to rapid Late CretaDEPAOLO, DJ. 1981. A neodymium and strontium isoceous batholith emplacement in the Peninsular topic study of the Mesozoic calc-alkaline granitic Ranges of Southern and Baja California. Geology, batholiths of the Sierra Nevada and Peninsular 29,491-494. Ranges, California. Journal of Geophysical LEWIS, J.L., DAY, S.M., MAGISTRALE, H., EAKINS, J. Research, 86, 10470-10488. & VERNON, F. 2000. Regional crustal thickness DILEK, Y. & MOORES, E.M. 1999. A Tibetan model for variations of the Peninsular Ranges, southern the Early Tertiary western United States. Journal of the Geological Society, London, 156, 929-941. California. Geology, 28, 303-306.

CORDILLERAN ARC AND OROGENIC PLATEAUX LIVACCARI, R.F. 1991. Role of crustal thickening and extensional collapse in the tectonic evolution of the Sevier-Laramide orogeny, western United States. Geology, 19, 1104-1107. MAGISTRALE, H. & SANDERS, C. 1995. P wave image of the Peninsular Ranges batholith, southern California. Geophysical Research Letters, 22, 2549-2552. MCKENZIE, D., NIMMO, F., JACKSON, J.A., GANS, P.B. & MILLER, E.L. 2000. Characteristics and consequences of flow in the lower crust. Journal of Geophysical Research, 105, 11029-11046. MCQUARRIE, N. & CHASE, C.G. 2000. Raising the Colorado Plateau. Geology, 28, 91-94. MILLER, R.B. & PATERSON, S.R. 2001. Influence of lithological heterogeneity, mechanical anisotropy, and magmatism on the rheology of an arc, North Cascades, Washington. Tectonophysics, 342, 351370. MOLNAR, P. & LYON-CAEN, H. 1988. Some simple physical aspects of the support, structure, and evolution of mountain belts. In: CLARK, S.P., Jr., BURCHFIEL, B.C. & SUPPE, J. (eds) Processes in Continental Lithospheric Deformation. Geological Society of America Special Paper, 218, 179-207. NORLANDER, B.H., WHITNEY, D.L., TEYSSIER, C. &

VANDERHAEGHE, O. 2002. High temperature decompression of the Shuswap metamorphic core complex, Canadian Cordillera. Lithos, 61, 103-125. PARRISH, R.R., CARR, S.D. & PARKINSON, D.L. 1988. Eocene extensional tectonics and geochronology of the southern Omineca Belt, British Columbia and Washington. Tectonics, 7, 181-212. PATERSON, S.R. & MILLER, R.B. 1998. Magma emplacement during arc-perpendicular shortening: an example from the Cascades crystalline core, Washington. Tectonics, 17, 571-586. PATERSON, S.R., TOBISCH, O.T. & VERNON, R.H. 1991. Emplacement and deformation of granitoids during volcanic arc construction in the Foothills Terrane, central Sierra Nevada, California. Tectonophysics, 191, 89-110. PICKETT, D.A. & SALEEBY, J.B. 1993. Thermobarometric constraints on the depth of exposure and conditions of plutonism and metamorphism at deep levels of the Sierra Nevada Batholith, Tehachapi Mountains, California. Journal of Geophysical Research, 98, 609-629. POTTER, C.J. SANFORD, W.E. et al 1986. COCORP deep seismic reflection traverse of the interior of the North American Cordillera, Washington and Idaho: implications for orogenic evolutions. Tectonics, 5, 1007-1025. SCHMIDT, K.L. 2000. Investigation of arc processes: relationships among deformation, magmatism, mountain building, and the role of crustal anisotropy in the evolution of the Peninsular Ranges batholith, Baja California. PhD thesis, University of Southern California, Los Angeles. SCHMIDT, K.L. & PATERSON, S.R. 2002. A doubly vergent fan structure in the Peninsular Ranges batholith: transpression or local complex flow

175

around a continental margin buttress? Tectonics, 21, (doi 10.1029/2001TC001353). SILVER, L.T. & CHAPPELL, B.W. 1988. The Peninsular Ranges Batholith: an insight into the evolution of the Cordilleran batholiths of southwestern North America. Transactions of the Royal Society of Edinburgh, 79, 105-121. SPEAR, F.S. & PARRISH, R.R. 1996. Petrology and cooling rates of the Valhalla Complex, British Columbia, Canada. Journal of Petrology, 37, 733-765. STOCK, J. & MOLNAR, P. 1988. Uncertainties and implications of the Late Cretaceous and Tertiary position of North America relative to the Farallon, Kula, and Pacific plates. Tectonics, 7, 1339-1384. STOWELL, H.H. & CRAWFORD, M.L. 2000. Metamorphic history of the Coast Mountains orogen, western British Columbia and southeastern Alaska. In: STOWELL, H.H. & MCCLELLAND, W.C. (eds) Tectonics of the Coast Mountains, Southeastern Alaska and British Columbia. Geological Society of America Special Paper, 343, 257-283. STOWELL, H.H. & McClELLAND, W.C. (eds) 2000. Tectonics of the Coast Mountains, Southeastern Alaska and British Columbia. Geological Society of America Special Paper, 343. TEYSSIER, C. & WHITNEY, D.L. 2002. Gneiss domes and orogeny. Geology, 30, 1139-1142. TODD, V.R., ERSKINE, B.G. & MORTON, D.M. 1988. Metamorphic and tectonic evolution of the northern Peninsular Ranges batholith. In: Ernst, W.G. (ed.) Metamorphism and Crustal Evolution of the Western United States, Rubey volume VII. Prentice-Hall, Englewood Cliffs, 894-937. VALLEY, P.M., WHITNEY, D.L., PATERSON, S.R., MILLER, R.B. & ALSLEBEN, H. 2003. Metamorphism of deepest exposed arc rocks in the Cretaceous to Paleogene Cascades belt, Washington: evidence for large-scale vertical motion in a continental arc. Journal of Metamorphic Geology, 21, 203—220. VANDERHAEGHE, O. & TEYSSIER, C. 2001. Partial melting and flow of orogens. Tectonophysics, 342,451-472. WERNICKE, B., CLAYTON, R.W. et al 1996. Origin of high mountains in the continents; the southern Sierra Nevada. Science, 71, 190-193. WHITNEY, D.L., MILLER, R.B. & PATERSON, S.R. 1999. P-Tevidence for the magnitude of syn-metamorphic vertical tectonic motion in a contractional orogen: the North Cascades-Coast Mountains Belt, northern U.S. and Canadian Cordillera. Journal of Metamorphic Geology, 17, 73-88. WOLFE, J.A., FOREST, C.E. & MOLNAR, P. 1998. Paleobotanical evidence of Eocene and Oligocene paleolatitudes in midlatitude western North America. Geological Society of America Bulletin, 110, 664-678. ZANDT, G., VELASCO, A.A. & BECK, S.L. 1994. Composition and thickness of the southern Altiplano crust, Bolivia. Geology, 22, 1003-1006.

This page intentionally left blank

Displacement transfer between thick- and thin-skinned decollement systems in the central North American Cordillera

w. c. MCCLELLAND & j. s. OLDOW Department of Geological Sciences, University of Idaho, Moscow, ID 83844-3022, USA (e-mail: [email protected]) Abstract: Late Cretaceous-Early Tertiary contractional deformation along the Cordilleran margin of North America is represented by two distinct styles of foreland deformation, thinskinned and thick-skinned, that primarily differ in depth to their respective basal decollements. Given the coeval nature of contraction in regions experiencing different styles of deformation, displacement on deep-level detachments associated with thick-skinned basement-cored uplifts in the southern Rocky Mountains was kinematically linked with displacement along shallow-level detachments in the southern Canadian Rockies. In the central North American Cordillera, the transition in foreland decollement depth was accommodated by a NW-trending oblique ramp system. The oblique ramp extended from the basementcored uplifts of Wyoming along the northern margin of the Idaho batholith to the Shuswap crustal duplex in southeastern British Columbia. While accommodating transfer between differing structural levels, the Late Cretaceous to Paleocene displacement on the oblique ramp produced uplift and exhumation of high-grade metamorphic and plutonic rocks of the Idaho batholith between the NW-striking Lewis and Clark line and the Orofino shear zone. This resulted in truncation of the transpressional western Idaho shear zone, and may have localized the site of Eocene extension in this portion of the Cordillera.

The Late Mesozoic and Early Tertiary accretionary and contractional history of the western North American Cordillera is one of the best documented in the world (e.g. Oldow et al 1989; Gabrielse & Yorath 1991; Burchfiel et al 1992). Within this context, it serves as an excellent example of tectonic processes active at convergent plate margins and yields important insight into the crustal architecture responsible for the development of kinematically linked structures emanating from decollement systems at different crustal levels. The timing and structural history of Cordilleran deformation, when viewed in the context of deep crustal seismic reflection profiles and regional restorable crustal sections, argues for lithospheric-scale kinematic coordination of structures developed in different parts of the orogen. This realization led to the orogenic float concept (Oldow et al 1989, 1990), which argues that diffuse transpressional deformation along convergent margins is linked from the plate boundary to the foreland by a basal decollement system (e.g. Oldow et al 1989; Cook & Varsek 1994). The basal decollement system ranges from a discrete fault surface recognized beneath the frontal foldand-thrust belts (e.g. Bally et al 1966) to a zone

of distributed shear below hinterland regions of the orogen (Oldow et al 1990). The depth of the basal decollement progressively increases toward the hinterland and must exceed the depths recorded by the highest-grade metamorphic tectonites exposed above thrust faults within the contractional orogen. As such, the basal decollement certainly extends into the lower crust and probably coincides with the profound change in mechanical properties across crust-mantle boundary (Carter & Tsenn 1987). The need for mechanical decoupling of the deformed crustal section from the underlying basement has long been recognized in collisional orogens (e.g. Ampferer 1906) and gained acceptance with the proliferation of deep crustal seismic reflection profiles across continental and ocean-continent collision zones (Cook et al 1988, 1999; Choukroune 1989; Costain et al 1989; Roure et al 1989; Blundell et al 1992; Clowes et al 1995). Geodynamic models support the contention that decollement systems stretch across the entire width of the orogen for both ocean-continent and continent-continent collision (e.g. Harry et al 1995; Pfifrher et al 2001). Harry et al (1995) explored the consequences of a mechanically stratified lithosphere on the

From: GROCOTT, J., MCCAFFREY, K. J. W., TAYLOR, G. & TIKOFF, B. (eds) 2004. Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 177-195. 0305-8719/047$ 15 © The Geological Society of London 2004.

178

w. c. MCCLELLAND ETAL.

structural evolution of an ocean-continent collisional orogen during margin-normal shortening. Their models predict the formation of decollement systems at mid- and lower crustal depths. Decollement systems nucleate along mechanical boundaries within the crust and provide successively increased decoupling between shallow and deep structural levels within the orogen. Strain within the lithospheric section is partitioned into upper crustal, lower crustal and mantle domains, but strain domains essentially experience the same degree of internal shortening. With increased depth, strain is distributed over successively wider regions, with the consequence that deep strain domains extend into the continental interior well beyond foreland limits of shallow structures. The implication is that, even though decollement systems segregate the locus of strain accumulation within the lithospheric section, they do not completely decouple the system and mechanical communication is preserved between crustal and mantle strain domains. The conclusions drawn from geodynamic models support mass balance and kinematic compatibility considerations developed in the orogenic float conceptualization (Oldow et al. 1990). Development of through-going crustal decollement systems is a reflection of different mechanisms of strain accumulation but does not relax the need to preserve volumetric and displacement balance. Specifically, marginnormal shortening within the hinterland and foreland of an orogen must be balanced by shortening of the entire lithospheric column (Oldow et al. 1990). The mechanism for lithospheric volume balance differs along the axis of the North American Cordillera and reflects the presence or absence of large-scale terrane accretion during continental shortening. In the Canadian Cordillera, a region characterized by terrane accretion (Coney et al. 1980), lithospheric balance is accommodated by removal of sub-decollement mantle lithosphere from beneath terranes and transfer of the terrane crustal sections to the continental orogen. Viewed as an end-member, orogen balance can be maintained by the addition of terranes in sufficient volume to accommodate shortening in the foreland. This contrasts sharply with regions lacking terrane accretion, where foreland shortening must be balanced by shortening or removal of sub-decollement lithospheric mantle (Oldow et al 1990). In both cases, contraction within the foreland occurs above a decollement that dips toward the plate boundary and extends below the orogen hinterland. Where sub-decollement lithospheric balance is accommodated is not well

understood, but it is probably concentrated along or near the plate boundary (Harry et al. 1995). Except in the case of complete delamination of the crustal section from the underlying lithospheric mantle, some degree of coupling is preserved across the basal decollement, and displacements above and below the decollement are kinematically linked as a means of maintaining lithospheric volume and displacement balance. In addition to mass balance, preservation of kinematic compatibility is a direct consequence of the orogenic float model, even where differing styles of structures accommodate crustal deformation along the axis of the orogen. This is true for the transition between thin- and thickskinned deformation in the central North American Cordillera (Armstrong 1968; Schmidt & Perry 1988), where the spatial distribution and timing of thin- and thick-skinned structures in foreland regions of the orogen are well established (see Allmendinger 1992). In the Canadian Rocky Mountains, thin-skinned foreland contraction migrated easterly through the Jurassic and Cretaceous (Bally et al. 1966; Price 1981) as terranes were accreted to the continental borderland (e.g. Monger et al. 1982, 1985), and thin-skinned shortening continued to the end of contraction in the Eocene (Price & Mountjoy 1970). Outboard of the southern Rocky Mountains, along the continental margin of the conterminous USA, no significant terrane accretion occurred in the Mesozoic; nevertheless, the Late Cretaceous (approximately 80 Ma) signalled a profound change in tectonic style with the transition from thin-skinned shortening of the Sevier foreland belt to thick-skinned basementcored Laramide uplifts (Armstrong 1968; Gries 1983). The Montana disturbed belt marks the transition between thin-skinned foreland structures of the southern Canadian Rockies and thick-skinned structures of the southern Rocky Mountains (Fig. 1). Deformation in the two structural domains was synchronous and records similar magnitudes of shortening (Wiltschko & Dorr 1983; Dorr et al. 1987; DeCelles et al. 1987; Dickinson et al. 1988; DeCelles & Mitra 1995). The primary difference between the domains stems from the variation in decollement level active during contraction. The nature and implications for the presence of this transition are explored below.

Crustal architecture of the central North American Cordillera The shallowly west-dipping basal decollement underlying the foreland fold-and-thrust belt of

DECOLLEMENT SYSTEMS

179

Fig. 1. Generalized tectonic map of the central North American Cordillera (modified after King 1969), showing the locations of cross-sections A-A' through H-H' presented in Figure 2. BT, Beartooth Mountains; CRE, Columbia River embayment; KC, Kettle extensional complex; LCL, Lewis and Clark lineament; MDB, Montana disturbed belt; OC, Okanagan extensional complex; OSZ, Orofino shear zone; PC, Priest River extensional complex; SC, Shuswap extensional complex; SRP, Snake River plain; WISZ, western Idaho shear zone; WR, Wind River Mountains.

the southern Canadian Rockies separates imbricated Phanerozoic and Proterozoic sedimentary rocks from the underlying Precambrian crystalline basement. Initial recognition of the decollement grew out of regional restorable sections (Bally etal 1966; Dahlstrom 1969; Price 1981) and gained widespread acceptance with the acquisition and publication of 'Lithoprobe' deep crustal seismic reflection profiles (Cook

et al 1988, 1992). Regional restorable sections across the southern Canadian Rockies, although varying in detail, consistently record shortening of between 150 and 250 km (Bally et al 1966; Price 1981), and uniformly depict the basal decollement extending from shallow levels within the Phanerozoic sedimentary succession of the Alberta foothills to depths of 25 km in the hinterland of eastern British Columbia and

180

w. c. MCCLELLAND ETAL

near the Idaho—Montana border. Our regional sections (Fig. 2a & b), taken from the work of Norris & Bally (1972) and Bally (1984), are consistent with Lithoprobe reflection profiles across the southern Canadian Rockies (Cook et al 1988, 1992; Cook 1995a). Although the structural framework of the foreland fold-and-thrust belt is well understood, uncertainty persists about how the foreland decollement relates to structures of the Shuswap complex in southeastern British Columbia (Fig. 1). Large-magnitude Eocene extension (Parrish et al 1988) reorganized the geometry of earlier contractional structures and was accompanied by a shoaling of the moho beneath the metamorphic hinterland (Cook et al. 1992; Cook 1995 b). Many deep crustal reflectors imaged by Lithoprobe beneath the Shuswap complex can be interpreted in both a contractional and an extensional context (Cook 1995c). As a result, the tectonic evolution of the metamorphic tectonites of the Shuswap complex and their relation to the foreland fold-and-thrust belt is based primarily on structural, metamorphic and geochronological investigations (e.g. Brown et al 1992; Carr 1995; Parrish 1995). It is generally accepted that the metamorphic tectonites in the Shuswap complex were exhumed and exposed by west- and east-dipping Eocene extensional fault systems that bound the metamorphic core (Brown & Journeay 1987; Parrish etal 1988; Brown et al 1992; Cooker al 1992). In the Shuswap region, the Monashee complex and lower parts of the overlying Selkirk allochthon contain numerous shear zones that progressively decrease in age from Early Cretaceous to Paleocene with depth (Parrish 1995). The shear zones accommodated ENE-directed transport of Proterozoic to Triassic rocks over autochthonous North American basement in Late Cretaceous to Paleocene time (e.g. Brown et al 1992; Carr 1995; Parrish 1995). Although basement extrusion (Johnson et al 2000) or flow (Vanderhaeghe SL Teyssier 2001) models have been proposed for the dynamics of basement exhumation, the early contractional history was clearly dominated by the formation of a crustal duplex (Brown et al 1986, 1992; Carr 1995; Parrish 1995) in the hinterland of the foreland fold-and-thrust belt. Although now obscured by later extensional structures and uplift of the moho, the Shuswap duplex predictably was underlain by a frontal ramp that migrated toward the foreland and duplex horses progressively transferred basement rocks from the footwall to hanging wall assemblage of the orogen. The structural style of the Rocky Mountain orogen undergoes a profound change to the south

across the US-Canadian border. Thin-skinned structures of the Alberta foreland give way to basement-cored uplifts that constitute the classic thick-skinned Laramide foreland structures of the southern Rocky Mountains. Specifically, in central Montana, displacement on thin-skinned thrusts is transferred to faults bounding basement uplifts of the Laramide foreland and accommodated by displacement on structures with WNW-trending axes (Fig. 1). The axes change to NNW trends in northern and central Wyoming. The switch from thin- to thickskinned deformation is accompanied by a southerly decrease in shortening on thin-skinned structures to approximately 40 km in northwestern Montana (Price & Sears 2000). Shortening by Laramide foreland structures is estimated at 55 km (Gries 1983; Allmendinger 1992) and is accommodated by broad crustal-scale folds and displacement on variably oriented thrust faults. The thrusts bounding the basement uplifts are steeply dipping and locally are traced in seismic reflection profiles to depths of 35 km and possibly deeper (Berg 1962; Smithson et al 1978; Sharry et al 1986). As the faults extend to depth, they broaden into distributed shear zones that lose definition in the lower crust. Nevertheless, the underlying moho is not offset by the structures (Smithson et al 1978, 1979; Sharry et al 1986), pointing to partial decoupling either in the lower crust or at the crust-mantle boundary (Oldow et al 1989, 1990). West of the Laramide foreland, in Idaho and western Montana, thin-skinned structures are preserved and form the frontal Sevier foldand-thrust belt (Fig. 1; Royse et al 1975). The Sevier fold-and-thrust belt was active from the Mid-Jurassic through the Cretaceous (Royse et al 1975; Villein & Kligfield 1986), accommodated between about 125 and 150 km of shortening (Royse etal 1975; Allmendinger 1992), and has a structural style like that of the southern Canadian foreland. Local crosscutting relations amongst Sevier thrusts and Laramide-style uplifts (Wiltschko & Dorr 1983; Dorr et al 1987), together with regional sedimentary dispersal patterns (DeCelles et al 1987; Dickinsonet al 1988), indicate that both thin- and thickskinned structures were active in the latest Cretaceous and earliest Paleocene. In the Early Tertiary, displacement in the Sevier belt ceased and migrated into the foreland, where deformation was dominated by displacement on deep decollement systems (e.g. Oldow et al 1989; Burchfiel et al 1992). Regional timing relations support the contention that the Sevier and Laramide belts are kinematically linked (e.g. Oldow et al 1989) by

Fig. 2. Restorable cross-sections of the central North American Cordillera, including: ENE- WSW sections across (a) Alberta-British Columbia (modified after Norris and Bally 1972), (b) Alberta-Idaho (modified after Bally 1984), (c) Montana-Idaho and (d) Wyoming-Idaho (both modified after Bally & Snelson 1980); and north-south sections across (e) Montana-Wyoming, (f) Montana-eastern Idaho, (g) Montana-central Idaho and (h) Montana-western Idaho.

Fig. 2. Continued.

DECOLLEMENT SYSTEMS

a structural architecture represented in regional sections across the Laramide foreland into the Sevier hinterland (Fig. 2d). Coeval east- and west-vergent Laramide foreland structures are interpreted to root into a basal decollement at depths of perhaps 40 km. The basal decollement is inferred to project west beneath the thinskinned Sevier belt to depths of 55 km beneath south-central Idaho (Fig. 2d). Thin-skinned foreland thrusting in the southern Canadian Rockies was demonstrably active in the Late Cretaceous and Early Tertiary (e.g. Price & Mountjoy 1970; Brown et al 1986; Fillipone & Yin 1994; Sears 2001). Similarly, basement-cored uplifts characteristic of Laramide deformation were active in latest Late Cretaceous through Early Tertiary (DeCelles et al. 1987; Dickinson et al 1988), overlapping in time with the late stages of thin-skinned deformation within the Sevier belt farther to the west (e.g. DeCelles & Mitra 1995). These timing relationships require coeval displacements on the shallow and deep basal decollement systems underlying the southern Canadian Rockies and the southern Rocky Mountains of the US Cordillera, respectively. To preserve kinematic compatibility between the northern and southern segments of the foreland thrust belt, displacements at shallow crustal levels in the Canadian foreland and at deep structural levels in the Shuswap complex (Fig. 2a & b) must be consistent with displacements on the deep and shallow decollement systems of the Laramide-Sevier orogen farther south (Fig. 2c & d). As a consequence, the decollement systems must be linked by a crustal ramp in the basal decollement of the orogen. From north to south along the foreland, the shallow basal decollement of the Canadian Rockies must step down to lower crustal depths beneath the Laramide foreland. The south-facing ramp has a structural height of 30-45 km in the foreland but decreases in height farther west in the hinterland as depicted in orogen-parallel sections (Fig. 2e-h). Essentially, the west-facing frontal ramp of the Shuswap crustal duplex is kinematically linked to Laramide foreland structures by an oblique crustal ramp system stretching across northern Idaho and southern Montana. Oblique ramp system in the Idaho-Montana basement The transition from thin- to thick-skinned structures in the northern USA is coincident with a number of other geological features that remain enigmatic with regard to their significance to

183

the evolution of the continental margin. The fold-and-thrust belt of southern Alberta and British Columbia extends southward to northern Idaho and Montana, where it involves the 15 to 21 km thick metasedimentary sequence of the Proterozoic Belt Supergroup (Winston 1986). Continuity of the dominantly north-trending structures of the fold-and-thrust belt is disrupted by a zone referred to as the Lewis and Clark line or lineament (Fig. 1; Billingsley & Locke 1939; Harrison et al. 1986). South of the Lewis and Clark line, the Belt Supergroup and its underlying basement are intruded by Cretaceous and Tertiary plutons that comprise the Bitterroot lobe of the Idaho batholith (Hyndman et al. 1988). The Lewis and Clark line is 40 to 80 km wide and defined by a complex array of NWstriking brittle faults as well as a series of NW-trending folds (Fig 1 & 3). On the basis of reported stratigraphic variations within the Belt Supergroup, the Lewis and Clark line is interpreted as the southern margin of the Proterozoic Belt basin (e.g. Harrison et al. 1974; White 1998). The zone is also inferred to be a deeplevel basin-margin ramp developed along the northern flank of the basin (Sears 1988; Price & Sears 2000). The lineament projects westward along the northern margin of the Columbia River embayment and eastward along the northern limit of the Montana disturbed belt (Fig. 1). We propose that the structural fabrics and distribution of metamorphic rocks along this structural trend reflect the surface manifestation of the crustal-scale ramp required to link thin-skinned and thick-skinned deformation to the north and south, respectively. The specific location and geometry of structures above the inferred ramp are derived from geological relations along the northern margin of the Idaho batholith (Fig. 3a). North of the Idaho batholith, north-trending contractional structures of the Rocky Mountain fold-and-thrust belt are overprinted by NWtrending folds within the Lewis and Clark line (Fig. 3a; Harrison et al. 1986; White 1998). The folds are in turn overprinted by high-angle faults that dominantly record dextral strike-slip displacement (Hobbs et al. 1965; Wallace et al. 1990). Although the lineament reportedly accommodated deformation in Precambrian time (e.g. Harrison et al. 1974), most relations suggest a Late Cretaceous and younger age for deformation within this zone (e.g. Wallace et al. 1990; Fillipone & Yin 1994). As discussed below, development of NW-striking folds in the Lewis and Clark line records deformation associated with Late Cretaceous displacement on the ramp structure. Displacement on faults associated with the Lewis and Clark line

184

w. c. MCCLELLAND ETAL

Fig. 3. (a) Generalized geological map of northern Idaho-western Montana (modified after Lewis 1995). (b) Schematic cross-section depicting geometry of proposed basement ramp between the Lewis and Clark line and Orofino shear zone. PC, Priest River extensional complex; BC, Bitterroot extensional complex.

DECOLLEMENT SYSTEMS

may have in part been synchronous with fold development, but most of the displacement is likely Tertiary and younger in age (Hobbs et al, 1965; Fillipone et al 1995). South of the Lewis and Clark line, metasedimentary rocks of the Belt Supergroup change from sub-greenschist to amphibolite facies (Hyndman et al. 1988). The higher-grade rocks in the Snow Peak area and other areas to the east (Fig. 3a) record a polymetamorphic history with peak pressures of approximately 0.6 GPa (Lang & Rice 19850, b). This assemblage is thrust southward over higher-grade rocks (Lewis 1995) that yield peak pressures in 0.8-1.0 GPa range (Carey et al. 1992; Grover et al. 1992; House et al. 1997; Foster et al. 2001). Available geochronological and thermochronological data from this portion of the Idaho batholith indicate that metamorphism of this higher-grade section together with the emplacement of deep-level plutons of the northern Idaho batholith occurred at mid- to lower crustal levels during Late Cretaceous time, at approximately 75 Ma (Toth & Stacey 1992). Younger phases of the Idaho batholith were emplaced into this section at increasingly shallower crustal levels between 65 and 55 Ma, indicating that deep-level rocks of the Idaho batholith were in part exhumed during Late Cretaceous time (Foster & Fanning 1997; House et al. 1997; Foster et al. 2001). Continued exhumation of the eastern portion of the Idaho-Bitterroot batholith was accommodated by Eocene extension in the Bitterroot extensional complex (Fig. 3; Hyndman 1980; Doughty & Sheriff 1992; House et al. 1991 \ Foster et al. 2001). High-grade metamorphic rocks within the Idaho-Bitterroot batholith continue southward to the Orofino area where kyanite/sillimanite-bearing upper amphibolite facies metamorphic rocks lie in the hanging wall of a SW-vergent shear zone (Hietanen 1962; Strayer et al. 1989), referred to as the Orofino shear zone (Fig. 3a; Payne & McClelland 2002). In the Orofino area, this structure is roughly coincident with the 0.706 and 0.704 initial Sr isopleths (Armstrong et al. 1911 \ Criss & Fleck 1987) and has been viewed as the northern extension of the western Idaho shear zone (Strayer et al. 1989), an early Late Cretaceous structure that separates metamorphic rocks of continental affinity from accreted island arc material of the Wallowa terrane to the west (Fig. 3a; Lund & Snee 1988; Manduca et al. 1993). However, structural and timing relationships suggest that the Orofino shear zone truncates the western Idaho shear zone (Payne et al. 2001). Foliations and lineations within the western Idaho shear zone are

185

traced to SE of Orofino, where they are disrupted and overprinted by structures of the Orofino shear zone. This younger structure is coincident with the Trans-Idaho discontinuity (Yates 1968) and can be traced southeastward past the northern projection of the western Idaho shear zone as a broad zone of deformation characterized by ductile fabrics. The ductile fabrics are overprinted by brittle structures of the Glade Creek fault zone (Payne et al. 2001), and timing constraints on the structure provided by 40Ar/39Ar (Snee et al. 1995) and U-Pb ages (Payne & McClelland 2002) suggest that displacement within the ductile shear zone occurred between 80 and 60 Ma. Exhumation of deep-level rocks within the northern Idaho batholith during the Late Cretaceous is a predicted consequence of displacement above an oblique crustal ramp linking deep decollement levels and shallower decollement systems related with the thin-skinned foreland deformation to the NE (Fig. 3b). In this oblique ramp model, high-grade rocks are inferred to have wedged beneath metasedimentary thrust sheets at shallow structural levels in a triangle zone geometry, the tip of which is coincident with the Lewis and Clark line (Fig. 3b). Rocks metamorphosed at intermediate crustal levels in the Snow Peak area were carried NE over the crustal ramp to the shallow flat underlying the frontal part of the southern Canadian Rockies. The Orofino shear zone is interpreted as a backthrust that developed above the crustal-scale ramp in the basal decollement (Fig. 3b). Assuming that NE-SW upper-plate transport directions for both thin- and thick-skinned displacements are similar, the ramp structure represents an oblique ramp in the central North American Cordillera. The location and orientation of the proposed ramp structure are inferred to reflect control by pre-existing basement structures. The exact nature of these structures is unknown but they most likely either reflect local structural trends within Archean basement or were formed during the Proterozoic rifting of the western North American margin (e.g. Price & Sears 2000).

Discussion Kinematic linkage between deep and shallow expressions of the Late Mesozoic and Tertiary foreland basal decollement is required to maintain volume balance and displacement field compatibility in the central North American Cordillera (Fig. 4a). Transitions between shallow and deep parts of the regional decollement vary across-strike between the Alberta foreland and Shuswap complex in the southern Canadian

186

w. c. MCCLELLAND ETAL

Fig. 4. (a) Late Cretaceous-Tertiary structural elements of the central North American Cordillera. CRE, Columbia River embayment; CSZ, Coast shear zone; LCL, Lewis and Clark lineament; OSZ, Orofino shear zone; PI, Pinchi fault; TI, Tintina fault; WISZ, western Idaho shear zone; WNSZ, western Nevada shear zone. Shaded area represents oblique ramp system between the LCL and OSZ connecting the Shuswap complex and region of thick-skinned uplifts, (b) Structural contour diagram of the basal decollement showing the location of the proposed ramp structure (solid lines) and the backthrust (dashed lines) associated with it. (c) Distribution of mid-Tertiary extensional complexes relative to the oblique ramp structure. BC, Bitterroot complex; FSC, Fraser-Straight Creek fault; SC, Shuswap complex; see Fig. 4a for other abbreviations.

DECOLLEMENT SYSTEMS

187

Fig. 4. Continued.

Rockies and along-strike between the Alberta foreland and the Laramide structures of the southern Rocky Mountains. Within this context, the depth to the basal decollement of the orogen is deflected by crustal-scale ramps. The Shuswap duplex migrated up a frontal ramp system during progressive Late Cretaceous to Tertiary uplift of metamorphic tectonites in the hinterland of the southern Canadian Rockies. Similarly, we argue that uplift of metamorphic tectonites around the Bitterroot lobe of the Idaho batholith occurred on an oblique ramp system stretching from the southern end of the Shuswap complex to the northern terminus of Laramide structures in the southern Rocky Mountain foreland. The location of an oblique ramp within the basal decollement system is expected to play an important role in the evolution of regionally significant structures. Regional expression of the oblique ramp To illustrate our view of the three-dimensional geometry of the regional decollement system beneath the central North American Cordillera, we present a structural contour map of both the east-directed decollement system and westdirected backthrust system (Fig. 4b). Depth estimates are based on interpolation between the regional sections shown in Figure 2 and the distribution of metamorphic rocks and present-day

crustal thickness estimates derived from sparse moho depth determinations (Mooney & Weaver 1989; Pakiser 1989; Prodehl & Lipman 1989). At the regional scale, the area underlain by relatively high-grade gneisses of the IdahoBitterroot batholith is bound by the Lewis and Clark line to the north and the Orofino shear zone to the south (Figs 3a & 4a). We argue that the uplift and exposure of these metamorphic rocks correspond to the subsurface position of an oblique ramp in the basal decollement system. The expression of upper plate deformation during NE-directed displacement over the basal ramp resulted in the formation of a rampbackthrust structure (Fig. 3b). The oblique ramp system projects southeastward into the Montana disturbed belt, where a broad antiformal uplift cored by Archaean rocks distorts older thrust sheets carrying Mesozoic to Precambrian sedimentary rocks (Fig. 1). Farther SE, the ramp-backthrust geometry is reflected in the southeastern continuation of the ramp where it merges with the frontal thrust of the thickskinned structures in the Beartooth uplift and the backthrust expressed by the NE-dipping thrust along the flank of the Wind River uplift (Figs 1 & 4a). Continuing southward, the frontal thrustbackthrust pair can be followed into the central and southern Rocky Mountain regions of Wyoming and Colorado (Fig. 4a; Erslev 1993).

188

w. c. MCCLELLAND ETAL

Assessing the westward continuation of the crustal ramp from northern Idaho is more problematic due to younger extension-related deformation and extensive Miocene volcanic cover in the Columbia River embayment. However, the ramp structure is inferred to turn northward along the western margin of the Shuswap complex (Figs 1 & 4a). The presence of a basement ramp in this region was proposed previously to explain older east- and west-vergent structures in the fold-and-thrust belt that are geometrically similar to those described here (Price 1986; Cook et al 1992; Colpron et al 1998). Early Tertiary migration of deep-level rocks up this crustal ramp may have triggered the onset of extensional deformation within the Shuswap-OkanaganKettle-Priest River extensional complex at approximately 58 Ma (Fig. 1; Carr 1992; Brown et al. 1992; Johnson & Brown 1996; Vanderhaeghe et al. 1999). Observed shortening on the Shuswap duplex progressively decreases to the north. The displacement is assumed to be transferred westward to structures such as the Coast shear zone or other similar transpressional structures active along the Late CretaceousEarly Tertiary Cordilleran margin. Influence on Late Cretaceous-Early Tertiary strike-slip systems of the Cordillera Late Cretaceous and Early Tertiary contractional deformation in the North American Cordillera is roughly contemporaneous with dextral displacement along several orogen-parallel transcurrent faults (e.g. Price & Carmichael 1986). These structures include the Tintina fault Pinchi fault, and Coast shear zone, but their southern extension into the US Cordillera is uncertain (Fig. 4a). Nevertheless, substantial plate-boundary-parallel displacement is required by geologically and palaeomagnetically determined estimates of 500-3000 km of Late Cretaceous and Tertiary northward displacement for outboard terranes (e.g. Baja British Columbia). Given the magnitudes of coeval shortening in the foreland and strike-slip displacement on the margin-parallel structures, this system of transcurrent structures was kinematically linked with the hinterland projections of the basal decollements of the thick- and thin-skinned fault systems (Oldow et al. 1989). Likewise, constraints of kinematic compatibility apply to strike-slip displacement on older structures such as the western Nevada shear zone (Wyld & Wright 2001) and the western Idaho shear zone (McClelland et al 2000). These structures may have accommodated some margin-parallel displacement of outboard terranes

(Fig. 4a), but continued lateral displacement on these structures is incompatible with development of the Idaho-Montana oblique ramp in the Late Cretaceous to Eocene. Structures associated with the oblique ramp truncate the western Idaho suture zone and accommodate the uplift of the northern Idaho batholith in the vicinity of the basement ramp. Specifically, the backthrust structure represented by the Orofino shear zone ostensibly truncated and offset the western Idaho shear zone. The geometry of this truncation is consistent with the apparent westward bend currently observed in the shear zone as defined by Sr isopleths (Armstrong et al. 1911 \ Criss & Fleck 1987). Interaction of the western Idaho shear zone and basement ramp is inferred to have terminated strike-slip displacement along this particular segment of the transpressional structure. This displacement was likely transferred via unrecognized structures in the Columbia River embayment to similar structures farther west, such as the Coast shear zone (McClelland et al 2000; Payne et al 2001). Although strike-slip displacement was terminated by development of the Orofino shear zone, the western Idaho shear zone continued to accommodate vertical displacement associated with Late Cretaceous and possibly Early Tertiary exhumation of the Idaho batholith (Snee et al 1995). Following its truncation, the western Idaho shear zone was translated approximately 100km to the NE relative to its deep-level counterpart in the lithospheric mantle (Leeman et al 1992). This model suggests that strikeslip systems in the Cordillera likely consisted of multiple strands operating at multiple structural levels throughout the evolution of the Rocky Mountain foreland fold-and-thrust belt. Influence on Tertiary extensional systems The Lewis and Clark line, which is coincident with the northernmost expression of the NWtrending basement ramp in northern Idaho— western Montana, is in large part defined by Tertiary and younger strike-slip faults. These strike-slip structures have been proposed as transfer structures between the Shuswap-OkanaganKettle-Priest River and Bitterroot extensional complexes (Figs 3 & 4c; Rehrig et al. 1987; Hyndman et al. 1988). Although this interpretation has been called into question on the basis of available fault kinematic data (Fillipone et al. 1995), a transfer of extensional displacement observed in regions to the north and south of the Lewis and Clark line is nevertheless expected. Estimates of extension for the Shuswap Okanagan-Kettle-Priest River and Bitterroot

DECOLLEMENT SYSTEMS

complexes are 100km and 50km, respectively (Parrish et al 1988; Hyndman 1980). The discrepancy implies that not all of the extension was transferred along the Lewis and Clark line. The excess is most likely accommodated within the Columbia River embayment south of the Okanagan-Kettle-Priest River complex (Fig. 4c). A similar transition is expected at the southern end of the Bitterroot complex, which lies along the southeastern projection of the Orofino shear zone (Fig. 4c). Extensional displacement is expected to be kinematically linked with coeval strike-slip displacement along outboard structures such as the Fraser-Straight Creek fault system (Fig. 4c; Price & Carmichael 1986). Kinematic model for the linked decollement system Late Cretaceous through Early Tertiary deformation of the central North American Cordillera involves substantial transcurrent and contractional displacement (Oldow et al. 1989) with the shortening component of displacement accommodated in a doubly vergent orogen (e.g. Oldow et al. 1989). As a first approximation, three-dimensional displacements within the orogenic system can be modelled by the displacement of blocks above a basal decollement that stretches across the width of the Cordillera.

189

In Figure 5 we show a map view of the contractional component of the displacement field for the central North American Cordillera. The boundary-parallel component of displacement associated with regional transcurrent faults is poorly understood for the Late Cretaceous to Early Tertiary time interval and is shown in the model as motion on a steeply dipping fault system. The magnitude of strike-slip displacement is controversial for the entire region as well as on individual segments of the fault system. The location of major through-going transcurrent faults is reasonably well established for the Canadian Cordillera, but the locus of transcurrent displacement in the northwestern US Cordillera is poorly understood. This ambiguity in transcurrent displacement adds additional uncertainty to the model, particularly in the relative location of the west-vergent belt of shortening represented by the Coast shear zone and its spatial relationship to contractional structures of the orogen farther east. Fortunately, the geometric constraint of a shared decollement linking transcurrent and contractional structures within the orogen mitigates the problem, such that it does not severely limit conclusions drawn about the contractional history of motion. In the model, we have rectilinearized the major contractional elements of the orogen. Eastdirected structures are depicted as shaded panels, where the dashed line marks the initial Late

Fig. 5. Late Cretaceous-Early Tertiary kinematic model for thin-skinned, thick-skinned and west-vergent backthrust structural elements of the central North American Cordillera. See text for discussion.

190

w. c. MCCLELLAND ETAL

Cretaceous location of the thrust front and the solid line shows the final location after displacement, which is parallel to the arrows. The lightgrey region corresponds to the thin-skinned foreland thrust belt of the Canadian Rockies. The Shuswap complex and east-directed structures of the Laramide foreland are shown in dark grey and linked by the ESE-trending oblique ramp system discussed in this chapter. West-directed structures are shown as panels without shading. Here also the initial location of the structure is shown as a dashed line and the final location as a solid line, with the displacement reflected by the arrows. During progressive deformation within a bi-vergent orogen, the location of westvergent structures must move in concert with east-directed contraction such that the entire orogen migrates toward the continental interior. Where easterly displacement of west-vergent structures does not keep pace with the foreland encroachment, orogen-normal extension results, but is not considered in this model. The magnitude of shortening across west-vergent structures is depicted by the unbound hachured areas, which represent the footwall area consumed. In this model, the total shortening across the orogen at any latitude is conserved. Thus both west- and east-vergent structures contribute to the total shortening budget, which does not vary along-strike. Possibly the simplest way to view the depiction is to image ENE displacement of the Pacific margin with respect to a fixed central North American continent. In this displacement scenario, some fractional component of the ENE motion of the Pacific coast is accommodated as the footwall is overridden by the contractional orogen (east-vergent structures). The displacement residual is taken up along west-vergent structures where the footwall slides beneath the orogen. In the northern Canadian Cordillera, orogenwide estimates of contraction must include eastvergent displacement within the thin-skinned thrust system of the foreland and a west-directed contractional component accommodated by the Coast shear zone of southeastern Alaska. Several tens of kilometres of shortening was accommodated by the Coast shear zone and was responsible for the uplift and exhumation of the Coast Mountains batholith in the Late Cretaceous and Early Tertiary (e.g. Crawford et al. 1987; McClelland & Mattinson 2000). The total shortening across the orogen is reflected by the easterly migration of the foreland thrust front and the attitudinally corresponding line length of western footwall consumed beneath the western margin of the orogen. The northward increase in shortening across the Coast shear

zone is depicted by a concomitant decrease in foreland shortening in this model, but could as easily be accommodated by internal shortening elsewhere in the orogen. Structures accommodating regional shortening in the southern Canadian and northernmost US Cordillera differ from the region farther north by the inclusion of the Shuswap complex. The Shuswap crustal duplex has a well-defined expression in the south but dies out to the north over a distance of about 1000 km. We illustrate the north to south increase in shortening across the Shuswap duplex as a corresponding decrease in the contraction accommodated by the Coast shear zone. At the southern end of the Shuswap virtually all of the hinterland shortening can be accommodated and a continued southern extent of the Coast shear zone is not required by, nor precluded by, the model. Displacement across the foreland belt is constant until the ENE projection (parallel to transport direction) of the southern end of the Shuswap is reached in western Montana. The stepped decrease in foreland shortening south of the projected end of the Shuswap is a geometric consequence of displacement transfer from the Shuswap to three belts of thick-skinned structures farther south. The dramatic decrease predicted by the geometric model is confirmed by estimates of foreland shortening in the Montana disturbed belt where thin-skinned contraction north of the Lewis and Clark line decrease abruptly (Price & Sears 2000). Shortening across the US Cordillera west of Laramide foreland is distributed across three belts of thick-skinned structures. In the east, shortening is accommodated by east- and west-vergent basement uplifts of the Laramide foreland. Farther west, we depict WSW-directed shortening across the northern Idaho-Montana basement ramp and by uplift of the western margin of the southern Idaho batholith on a successor structure to the western Idaho shear zone. This geometric exercise provides a simple test of the displacement compatibility proposed in our model of the central Cordillera. The model and relative displacements certainly are not unique, but they do provide a useful framework to explore the timing and magnitude of shortening in this region of the North American active margin. Coupling versus decoupling within the North American Cordillera The presence of a basement ramp in the basal decollement of the Cordilleran orogen is proposed to explain the observed transition between thin-skinned and thick-skinned foreland

DECOLLEMENT SYSTEMS

structures in the central Cordillera. The fundamental difference between the two regions is the depth to the decollement in the foreland. Decollement depths in the hinterland or west of the respective foreland domains are at similar structural levels at or near the base of the crust and only differ toward the frontal portions of the orogen. The existence of a deep-level detachment linking thick-skinned foreland structures of the southern Rocky Mountains to the hinterland is compatible with available seismic profiles. Deep crustal reflection profiles across Wind River Range of the southern Rocky Mountains document downward continuation of moderately dipping thrust faults to depths of at least 35 km (Smithson et al 1978, 1979) and are consistent with depths of up to 40 km. The moho beneath the Wind River uplift is not offset by ~12 km of vertical displacement observed at the surface and near-subsurface (Smithson et al. 1979) and mandates the existence of a deep decollement to decouple the crust and mantle. We do not promote the idea that the decollement is necessarily an abrupt structural discontinuity. Rather, at these depths we visualize the decollement as a zone of distributed shear that accommodates differential shortening associated with the differences in rheological properties of the crust and mantle. In light of known coeval and kinematically coordinated displacement of Sevier and Laramide structures in the Late Cretaceous, linkage is needed to transfer displacement from shallow to deep crustal levels in the foreland. Furthermore, we contend that the deep decollement beneath the Laramide foreland extended west toward the continental margin in regions where the structure now has been obliterated by uplift of the mantle during Late Cenozoic extension (Oldow et al. 1989). By proposing an orogen-extensive basal decollement, we are not implying complete decoupling of the crust and lithospheric mantle. Rather, as pointed out by Harry et al. (1995), a substantial degree of mechanical coupling is predicted from geodynamic models and required to accommodate displacement and volume balance considerations for lithospheric-scale deformation (Oldow et al. 1990). In this context, the orogenic float conceptualization is compatible with formation of Laramide uplifts in response to a basal traction (i.e. partial coupling or attachment) developed during differential shortening between a partially decoupled mantle and crust.

Conclusions Late Cretaceous to Eocene contraction in the central Rocky Mountain foreland of the North

191

American Cordillera is accommodated by both thin- and thick-skinned structures. Thin-skinned foreland structures of the southern Canadian Rockies are kinematically linked both to the Shuswap crustal duplex in southeastern British Columbia and to Laramide structures in the southern Rocky Mountain foreland of the US Cordillera. The north to south transition from thin- to thick-skinned foreland deformation is accommodated by a steep SW-facing crustal ramp that linked the shallow decollement system of the southern Canadian Rockies and a deep decollement beneath the basement uplifts of the Laramide foreland. The structural ramp was oblique to ENE regional shortening and transferred deep crustal displacement of Laramide foreland structures to the Shuswap duplex in the hinterland of the southern Canadian Rockies. The oblique ramp extended from the foreland transition in south-central Montana NW to northern Idaho and has a surface manifestation in the belt of basement-cored uplifts in southwestern Montana and exposure of plutonic and high-grade metamorphic rocks of the Idaho batholith in north-central Idaho. The location of the oblique ramp system apparently was nucleated by pre-existing crustal boundaries and the ramp evolved as an expression of displacement transfer required to preserve volumetric and displacement balance within a transpressional orogen. Once formed, the transfer structures become part of and exert control on continued displacements within the orogen. Partial support for this contribution came from US National Science Foundation grant EAR-0003007 awarded to WCMcC. Helpful reviews were provided by B. Tikoff, C. Teyssier and an anonymous reviewer.

References ALLMENDINGER, R.W. 1992. Fold and thrust tectonics of the western United States exclusive of the accreted terranes. In: BURCHFIEL, B.C., LIPMAN, P.W. & ZOBACK, M.L. (eds) The Cordilleran Orogen: Conterminous U.S. The Geology of North America, G-3, Geological Society of America, Boulder, Co, 583-607. AMPFERER, O. 1906. Ueber das Bewegungsbild von Faltengebirgen. Jahrbuch der Kaiserlich Koniglichen Geologischen Reichsanstalt, 56, 539-622. ARMSTRONG, R.L. 1968. Sevier orogenic belt in Nevada and Utah. Geological Society of America Bulletin, 79, 429-458. ARMSTRONG, R.L., TAUBENECK, W.H. & HALES, P.O. 1977. Rb-Sr and K-Ar geochronometry of Mesozoic granitic rocks and their Sr isotopic composition, Oregon, Washington, and Idaho. Geological Society of America Bulletin, 88, 397-411.

192

w. c. MCCLELLAND ETAL

BALLY, A.W. 1984. Tectogenese et sismique reflexion. Bulletin Societe Geologique de France, 7, 279—285. BALLY, A.W. & SNELSON, S. 1980. Realms of subsidence. In: MIALL, A.D. (ed.) Facts and Principles of World Petroleum Occurrence. Canadian Society of Petroleum Geologists Memoir, 6, 9-75. BALLY, A.W., GORDY, P.L. & STEWART, G.A. 1966. Structure, seismic data and erogenic evolution of southern Canadian Rocky Mountains. Bulletin of Canadian Petroleum Geology, 14, 337—381. BERG, R.R. 1962. Mountain flank thrusting in Rocky Mountain foreland, Wyoming and Colorado. American Association of Petroleum Geologists Bulletin, 46, 2019-2032. BILLINGSLEY, P.R. & LOCKE, A. 1941. Structure of ore deposits in the continental framework. Transactions of the Society of Mining Engineers of American Institute of Mining, Metallurgical and Petroleum Engineers, 144, 9-59. BLUNDELL, D., FREEMAN, R. & MUELLER, S. 1992. A Continent Revealed: The European Geotraverse. Cambridge University Press, Cambridge. BROWN, R.L. & JOURNEAY, J.M. 1987. Tectonic denudation of the Shuswap metamorphic terrane of southeastern British Columbia. Geology, 15, 142-146. BROWN, R.L., JOURNEAY, J.M., LANE, L.S., MURPHY, D.C. & REES, CJ. 1986. Obduction, backfolding and piggyback thrusting in the metamorphic hinterland of the southeastern Canadian Cordillera. Journal of Structural Geology, 8, 255-268. BROWN, R.L., CARR, S.D., JOHNSON, B.J., COLEMAN, V.J., COOK, F.A. & VARSEK, J.L. 1992. The Monashee decollement of the southern Canadian Cordillera: a crustal-scale shear zone linking the Rocky Mountain foreland belt to lower crust beneath accreted terranes. In: MCKLAY, K.R. (ed.) Thrust Tectonics. Chapman & Hall, London, 357-364. BURCHFIEL, B.C., COWAN, D.S., & DAVIS, G.A. 1992. Tectonic overview of the Cordilleran Orogen in the Western United States. In: BURCHFIEL, B.C., LIPMAN, P.W. & ZOBACK, M.L. (eds) The Cordilleran Orogen: Conterminous U.S. The Geology of North America, G-3. Geological Society of America, Boulder, CO, 407-479. CAREY, J.W., RICE, J.M. & GROVER, T.W. 1992. Petrology of aluminous schist in the Boehls Butte region of northern Idaho; geologic history and aluminum-silicate phase relations. American Journal of Science, 292, 455-473. CARR, S.D. 1992. Tectonic setting and U-Pb geochronology of the early Tertiary Ladybird leucogranite suite, Thor-Odin-Pinnacles area, southern Omineca Belt, British Columbia. Tectonics, 11, 258-278. CARR, S.D. 1995. The southern Omineca Belt, British Columbia: new perspectives from the Lithoprobe Geoscience program. Canadian Journal of Earth Sciences, 32, 1720-1739. CARTER, N.L. & TSENN, M.C. 1987. Flow properties of continental lithosphere. Tectonophysics, 136, 27-63.

CHOUKROUNE, P. 1989. The ECORS Pyrenean deep seismic profile reflection data and the overall structure of an erogenic belt. Tectonics, 8, 23-39. CLOWES, R.M., ZELT, C.A., AMOR, J.R. & ELLIS, R.M. 1995. Lithospheric structure in the southern Canadian Cordillera from a network of seismic refraction lines. Canadian Journal of Earth Sciences, 32, 1485-1513. COLPRON, M., WARREN, M.J., & PRICE, R.A. 1998. Selkirk fan structure, southeastern Canadian cordillera: tectonic wedging against an inherited basement ramp. Geological Society of America Bulletin, 110, 1060-1074. CONEY, P.J., JONES, D.L. & MONGER, J.W.H. 1980. Cordilleran suspect terranes. Nature, 288, 329-333. COOK, F.A., (ed.) I995a. The southern Canadian Cordillera transect of Lithoprobe. Canadian Journal of Earth Sciences, 32, 1483-1824. COOK, F.A. I995b. The southern Canadian Cordillera transect of Lithoprobe: introduction. Canadian Journal of Earth Sciences, 32, 1483-1484. COOK, F.A. 1995c. The reflection Moho beneath the southern Canadian Cordillera. Canadian Journal of Earth Sciences, 32, 1520-1530. COOK, F.A. & VARSEK, J.L. 1994. Orogen-scale decollements. Reviews of Geophysics, 32, 37-60. COOK, F.A., GREEN, A.G., et al. 1988. LITHOPROBE seismic reflection structure of the southeastern Canadian Cordillera, Initial results. Tectonics, 7, 157-180. COOK, F.A., VAN DER VELDEN, A.J., HALL, K.W. & ROBERTS, B.J. 1999. Frozen subduction in Canada's Northwest Territories: Lithoprobe deep lithospheric reflection profiling of the western Canadian Shield. Tectonics, 18, 1-24. COOK, F.A., VARSEK, J.L., et al. 1992. LITHOPROBE crustal reflection cross-section of the southern Canadian CORDILLERA, I. Foreland thrust and fold belt to Fraser River fault. Tectonics, 11, 12-35. COSTAIN, J.K., HATCHER, R.D., JR. & CORUH, C. 1989. Appalachian ultra-deep core hole (ADCOH) project site investigation, regional seismic lines, and geologic interpretation. In: HATCHER, R.D., JR., THOMAS, W.A. & VIELE, G.W. (eds) The Appalachian-Ouachita Orogen in the United States. The Geology of North America, F-2. Geological Society of America, Boulder, CO. CRAWFORD, M.L., HOLLISTER, L.S. & WOODSWORTH, G.J. 1987. Crustal deformation and regional metamorphism across a terrane boundary, Coast plutonic complex, British Columbia. Tectonics, 6, 343-361. CRISS, R.E. & FLECK, R.J. 1987. Petrogenesis, geochronology, and hydrothermal systems of the northern Idaho batholith and adjacent areas based on 180/160, D/H, 87Sr/86Sr, K-Ar and40Ar/39Ar studies. In: VALLIER, T.L. & BROOKS, H.C. (eds) Geology of the Blue Mountains Region of Oregon, Idaho, and Washington: The Idaho Batholith and Its Border Zone. United States Geological Survey Professional Paper 1436, 95-138. DAHLSTROM, C.D.A. 1969. Balanced cross sections. Canadian Journal of Earth Sciences, 6, 743-757.

DECOLLEMENT SYSTEMS DECELLES, P.O. & MITRA, G. 1995. History of the Sevier orogenic wedge in terms of critical taper models, northeast Utah and southwest Wyoming. Geological Society of America Bulletin, 107, 454-462. DECELLES, P.O., TOLSON, R.B. et al. 1987. Laramide thrust-generated alluvial-fan sedimentation, Sphinx Conglomerate, southwestern Montana. American Association of Petroleum Geologists Bulletin, 71, 135-155. DICKINSON, W.R., KLUTE, M.A., HAYES, M.J., JANECKE, S.U., LUNDIN, E., MCKITTRICK, M.A. & OLIVARES, M.D. 1988. Paleogeographic and paleotectonic setting of Laramide sedimentary basins in the central Rocky Mountain region. Geological Society of America Bulletin, 100, 1023-1039. DORR, J.A., JR., SPEARING, D.R., STEIDTMAN, J.R., WILTSCHKO, D.V. & CRADDOCK, J.P. 1987. Hoback River Canyon, central western Wyoming. In: BEUS, S.S. (ed.) Rocky Mountain Section of the Geological Society of America. Geological Society of America Centennial Field Guide, 2, 197-200. DOUGHTY, P.T. & SHERIFF, S.D. 1992. Paleomagnetic evidence for en echelon crustal extension and crustal rotations in western Montana and Idaho. Tectonics, 11, 663-671. ERSLEV, E.A. 1993. Thrusts, back-thrusts, and detachment of Rocky Mountain foreland arches. In: SCHMIDT, C.J., CHASE, R.B. & ERSLEV, E.A. (eds) Laramide Basement Deformation in the Rocky Mountain Foreland of the Western United States. Geological Society of America Special Paper 280, 339-358. FILLIPONE, J.A. & YIN, A. 1994. Age and regional tectonic implications of Late Cretaceous thrusting and Eocene extension, Cabinet Mountains, Northwest Montana and northern Idaho. Geological Society of America Bulletin, 106, 1017-1032. FILLIPONE, J.A., YIN, A., HARRISON, T.M., GEHRELS, G., SMITH, M. & SAMPLE, J.C. 1995. Age and magnitude of dip-slip faulting deduced from differential cooling histories; an example from the Hope Fault, Northwest Montana. Journal of Geology, 103, 199-211. FOSTER, D.A. & FANNING, C.M. 1997. Geochronology of the northern Idaho Batholith and the Bitterroot metamorphic core complex; magmatism preceding and contemporaneous with extension. Geological Society of America Bulletin, 109, 379-394. FOSTER, D.A., SCHAFER, C., FANNING, C.M. & HYNDMAN, D.W. 2001. Relationships between crustal partial melting, plutonism, orogeny, and exhumation; Idaho-Bitterroot Batholith. Tectonophysics,342, 313-350. GABRIELSE, H. & YORATH, C.J., (eds) 1991. The Cordilleran Orogen: Canada. Geological Survey of Canada, Geology of Canada 4. GRIES, R. 1983. Oil and gas beneath Precambrian of foreland thrust plates in the Rocky Mountains. American Association of Petroleum Geologists Bulletin, 67, 1-28. GROVER, T.W., RICE, J.M. & CAREY, J.W. 1992. Petrology of aluminous schist in the Boehls Butte region of northern Idaho; phase equilibria and

193

P-T evolution. American Journal of Science, 292, 474-507. HARRISON, I.E., GRIGGS, A.B. & WELLS, J.D. 1974. Tectonic Features of the Precambrian Belt Basin and Their Influence on Post-Belt Structures. United States Geological Survey Professional Paper 866. HARRISON, I.E., GRIGGS, A.B. & WELLS, J.D. 1986. Geologic and structure maps of the Wallace 1° x 2° quadrangle. United States Geological Survey Montana Map 1-1509, 1:250,000. HARRY, D.L., OLDOW, J.S. & SAWYER, D.S. 1995. The growth of orogenic belts and the role of crustal heterogeneities in decollement tectonics. Geological Society of America Bulletin, 107, 1411-1426. HIETANEN, A. 1962. Metasomatic Metamorphism in Western Clearwater County, Idaho. United States Geological Survey Professional Paper 344-A. HOBBS, S.W., GRIGGS, A.B., WALLACE, R.E. & CAMPBELL, A.B. 1965. Geology of the Coeur d'Alene district, Shoshone County, Idaho. United States Geological Survey Professional Paper 478. HOUSE, M.A., HODGES, K.V. & BOWRING, S.A. 1997. Petrological and geochronological constraints on regional metamorphism along the northern border of the Bitterroot Batholith. Journal of Metamorphic Geology, 15, 753-764. HYNDMAN, D.W. 1980. Bitterroot dome-Sapphire tectonic block, an example of a plutonic-core gneiss dome complex with its detached suprastructure. In: CRITTENDEN, M.D., CONEY, PJ. & DAVIS, G.H. (eds) Cordilleran Metamorphic Core Complexes. Geological Society of America Memoir 153, 427-443. HYNDMAN, D.W., ALT, D. & SEARS, J.W. 1988. Post-Archean metamorphic and tectonic evolution of western Montana and northern Idaho. In: ERNST, W.G. (ed.) Metamorphism and Crustal Evolution of the Western United States. PrenticeHall, Englewood Cliffs, NJ, 332-361. JOHNSON, B.J. & BROWN, R.L. 1996. Crustal structure and early Tertiary extensional tectonics of the Omineca belt at 51 °N latitude, southern Canadian Cordillera. Canadian Journal of Earth Sciences, 33, 1596-1611. JOHNSTON, D.H., WILLIAMS, P.P., BROWN, R.L., CROWLEY, J.L. & CARR, S.D. 2000. Northeastward extrusion and extensional exhumation of crystalline rocks of the Monashee complex, southeastern Canadian Cordillera. Journal of Structural Geology, 22, 603-625. KING, P.B. 1969. Tectonic Map of North America. United States Geological Survey Map, 1:5000,000. LANG, H.M. & RICE, J.M. 1985a. Metamorphism of pelitic rocks in the Snow Peak area, northern Idaho; sequence of events and regional implications. Geological Society of America Bulletin, 96,731-736. LANG, H.M. & RICE, J.M. 1985£. Geothermometry, geobarometry and T-X (Fe-Mg) relations in metapelites, Snow Peak, northern Idaho. Journal of Petrology, 26, 889-924. LEEMAN, W.P., OLDOW, J.S. & HART, W.K. 1992. Lithosphere-scale thrusting in the Western U.S.

194

w. c. MCCLELLAND ETAL

Cordillera as constrained by Sr and Nd isotopic transitions in Neogene volcanic rocks. Geology, 20, 63-66. LEWIS, R.S. 1995. Geologic Map of the Belt Supergroup in North-Central Idaho and Western Montana. United States Geological Survey Open-File Report 95-513, 1:500,000. LUND, K. & SNEE, L.W. 1988. Metamorphism, structural development, and age of the continentisland arc juncture in west-central Idaho. In: ERNST, W.G. (ed.) Metamorphism and Crustal Evolution of the Western United States. PrenticeHall, Englewood Cliffs, NJ, 296-331. MANDUCA, C.A., KUNTZ, M.A. & SILVER, L.T. 1993. Emplacement and deformation history of the western margin of the Idaho batholith near McCall, Idaho: influence of a major terrane boundary. Geological Society of America Bulletin, 105, 749-765. MCCLELLAND, W.C. & MATTINSON, J.M. 2000. Cretaceous-Tertiary evolution of the western Coast Mountains, central southeastern Alaska. In: STOWELL, H.H. & MCCLELLAND, W.C. (eds) Tectonics of the Coast Mountains Orogen. Geological Society of America Special Paper 343, 159-182. MCCLELLAND, W.C., TIKOFF, B. & MANDUCA, C.A. 2000. The role of intra-arc shear zones in terrane accretion: examples from the North American Cordillera. Tectonophysics, 326, 37-55. MONGER, J.W.H., PRICE, R.A. & TEMPLEMAN KLUIT, D.J. 1982. Tectonic accretion and the origin of two major metamorphic and plutonic welts in the Canadian Cordillera. Geology, 10, 70-75. MONGER, J.W.H., CLOWES, R.M., PRICE, R.A., SIMONY, P.S., RIDDIHOUGH, R.P. & WOODSWORTH, G.J. 1985. Centennial Continent/Ocean Transect #7, B-2 Juan de Fuca Plate to Alberta Plains. In: Speed, R. (ed.) Decade of North American Geology Continent/Ocean Transects. Geological Society of America, Boulder, CO. MOONEY, W.D. & WEAVER, C.S. 1989. Regional crustal structure and tectonics of the Pacific coastal states; California, Oregon, and Washington. In: PAKISER, L.C. & MOONEY, W.D. (eds) Geophysical Framework of the Continental United States. Geological Society of America Memoir, 172, 129-162. NORRIS, D.K. & BALLY, A.W. 1972. Coal, Oil, Gas and Industrial Mineral Deposits of the Interior Plains, Foothills and Rocky Mountains of Alberta and British Columbia, Guidebook. International Geological Congress Guidebook, 24, 106. OLDOW, J.S., BALLY, A.W., AVE LALLEMANT, H.G. & LEEMAN, W.P. 1989. Phanerozoic evolution of the North American Cordillera: United States and Canada. In: BALLY, A.W. & PALMER, A.R. (eds) The Geology of North America - An Overview. Geological Society of America, Boulder, CO, 139-232. OLDOW, J.S., BALLY, A.W. & AVE LALLEMANT, H.G. 1990. Transpression, orogenic float, and lithospheric balance. Geology, 18, 991-994. PAKISER, L.C. 1989. Geophysics of the Intermontane system. In: PAKISER, L.C. & MOONEY, W.D.

(eds) Geophysical Framework of the Continental United States. Geological Society of America Memoir, 172, 235-248. PARRISH, R.R. 1995. Thermal evolution of the southern Canadian Cordillera. Canadian Journal of Earth Sciences, 32, 1618-1642. PARRISH, R.R., CARR, S.D. & PARKINSON, D.L. 1988. Eocene extensional tectonics and geochronology of the southern Omineca Belt, British Columbia and Washington. Tectonics, 7, 181-212. PAYNE, J.D. & MCCLELLAND, W.C. 2002. Kinematic and temporal constraints for truncation of the western Idaho shear zone. Geological Society of America Abstracts with Programs, 34, A102. PAYNE, J.D., MCCLELLAND, W.C. & OLDOW, J.S. 2001. Truncation of the Western Idaho Suture Zone: kinematic constraints for post accretionary modification of the arc-continent boundary. American Geophysical Union EOS Transactions, 32, 167. PFIFFNER, O.A., ELLIS, S. & BEAUMONT, C. 2000. Collision tectonics in the Swiss Alps: insight from geodynamic modeling. Tectonics, 19, 1065-1094. PRICE, R.A. 1981. The Cordilleran thrust and fold belt in the southern Canadian Rocky Mountains. In: MCCLAY, K.R. & PRICE, NJ. (eds) Thrust and Nappe Tectonics. Geological Society, London, Special Publications, 9, 427-448. PRICE, R.A. 1986. The southeastern Canadian Cordillera: thrust faulting, tectonic wedging, and delamination of the lithosphere. Journal of Structural Geology, 8, 239-254. PRICE, R.A. & CARMICHAEL, D.M. 1986. Geometric test for Late Cretaceous-Paleogene intracontinental transform faulting in the Canadian Cordillera. Geology, 14, 468-471. PRICE, R.A. & MOUNTJOY, E.W. 1970. Geologic structure of the Canadian Rocky Mountains between Bow and Athabasca Rivers - a progress report. In: Wheeler, J.O. (ed.) Structure of the Southern Canadian Cordillera. Geological Association of Canada Special Paper, 6, 7-25. PRICE, R.A. & SEARS, J.W. 2000. A preliminary palinspastic map of the Mesoproterozoic Belt-Purcell Supergroup, Canada and USA: implications for the tectonic setting and structural evolution of the Purcell anticlinorium and the Sullivan deposit. In: LYDON, J.W., HOY, T., SLACK, J.F. & KNAPP, M.E. (eds) The Geological Environment of the Sullivan Deposit, British Columbia. Geological Association of Canada Mineral Deposits Division Special Publication, 1, 61-81. PRODEHL, C. & LIPMAN, P.W. 1989. Crustal structure of the Rocky Mountain region. In: PAKISER, L.C. & MOONEY, W.D. (eds) Geophysical Framework of the Continental United States. Geological Society of America Memoir, 172, 249-284. REHRIG, W.A., REYNOLDS, S.J. & ARMSTRONG, R.L. 1987. A tectonic and geochronologic overview of the Priest River crystalline complex, northeastern Washington and northern Idaho. Washington Division of Geology and Earth Resources Bulletin, 77, 185-193.

DECOLLEMENT SYSTEMS ROURE, R, CHOUKROUNE, P. etal. 1989. ECORS deep seismic data and balanced cross sections; geometric constraints on the evolution of the Pyrenees. Tectonics, 8, 41-50. ROYSE, F., JR., WARNER, M.A., & REESE, D.L. 1975. Thrust belt structural geometry and related stratigraphic problems, Wyoming, Idaho-northern Utah. In: BOLYARD, D.W. (ed.) Symposium on Deep Drilling in the Central Rocky Mountains. Rocky Mountain Association of Geologists, Denver, CO, 41-54. SCHMIDT, CJ. & PERRY, WJ. (eds) 1988. Interaction of the Rocky Mountain Foreland and the Cordilleran Thrust Belt. Geological Society of America Memoir, 171. SEARS, J.W. 1988. Two major thrust slabs in the westcentral Montana Cordillera. In: SCHMIDT, CJ. & PERRY, W.J. (eds) Interaction of the Rocky Mountain Foreland and the Cordilleran Thrust Belt. Geological Society of America Memoir, 171, 165-170. SEARS, J.W. 2001. Emplacement and denudation history of the Lewis-Eldorado-Hoadley thrust slab in the northern Montana Cordillera, USA: implications for steady-state orogenic processes. American Journal of Science, 301, 359-373. SHARRY, J., LANGAN, R.T., JOVANOVICH, D.B., JONES, G.M., HILL, N.R. & GUIDISH, T.M. 1986. Enhanced imaging of the COCORP seismic line, Wind River Mountains. In: BARAZANGI, M. & BROWN, L.D. (eds) Reflection Seismology; A Global Perspective. American Geophysical Union Geodynamics Series, 13, 223-236. SMITHSON, S.B., BREWER, J., KAUFMAN, S., OLIVER, J. & HURICH, C.A. 1978. Nature of the Wind River thrust, Wyoming, from COCORP deepreflection data and from gravity data. Geology, 6, 648-652. SMITHSON, S.B., BREWER, J., KAUFMAN, S., OLIVER, J.E., & HURICH, C.A. 1979. Structure of the Laramide Wind River uplift, Wyoming, from COCORP deep reflection data and from gravity data. Journal of Geophysical Research, 84, 5955-5972. SNEE, L.W., LUND, K., SUTTER, J.F., BALCER, D.E. & EVANS, K.V. 1995. An 40Ar/39Ar chronicle of the tectonic development of the Salmon River suture zone, western Idaho. In: VALLIER, T.L. & BROOKS, H.C. (eds) Geology of the Blue Mountains Region of Oregon, Idaho, and Washington: Petrology and Tectonic Evolution of Pre-Tertiary Rocks of the Blue Mountains Region. United States

195

Geological Survey Professional Paper, 1438, 359-414. STRAYER, L.M., HYNDMAN, D.W., SEARS, J.W. & MYERS, P.E. 1989. Direction and shear sense during suturing of the Seven Devils-Wallowa terrane against North America in western Idaho. Geology, 17, 1025-1029. TOTH, M.I. & STAGEY, J.S. 1992. Constraints on the Formation of the Bitterroot Lobe of the Idaho Batholith, Idaho and Montana, from U—Pb Zircon Geochronology and Feldspar Pb Isotopic Data. United States Geological Survey Bulletin, 2008. VANDERHAEGHE, O. & TEYSSIER, C. 2001. Partial melting and flow of orogens. Tectonophysics, 342,451-472. VANDERHAEGHE, O., TEYSSIER, C. & WYSOCZANSKI, R. 1999. Structural and geochronologic constraints on the role of partial melting during the formation of the Shuswap metamorphic core complex at the latitude of the Thor-Odin dome, British Columbia. Canadian Journal of Earth Science, 36, 917-943. VILLEIN, A. & KLIGFIELD, R. 1986. Thrusting and synorogenic sedimentation in central Utah. In: PETERSON, J.A. (ed.) Paleotectonics and Sedimentation in the Rocky Mountain Region, United States. American Association of Petroleum Geologists Memoir, 41, 281-307. WALLACE, C.A., LIDKE, D.J. & SCHMIDT, R.G. 1990. Faults of the central part of the Lewis and Clark line and fragmentation of the Late Cretaceous foreland basin in west-central Montana. Geological Society of America Bulletin, 102, 1021-1037. WHITE, B.G. 1998. Diverse Tectonism in the Coeur d'Alene Mining District. Montana Bureau of Mines and Geology Special Publication, 112, 254-265. WILTSCHKO, D.V. & DORR, J.A., JR. 1983. Timing of deformation in overthrust belt and foreland of Idaho, Wyoming and Utah. American Association of Petroleum Geologists Bulletin, 67,1304-1322. WINSTON, D. 1986. Belt Supergroup Stratigraphic Correlation Sections, Western Montana and Adjacent Areas. Montana Bureau of Mines and Geology Special Paper, 94. WYLD, S.J. & WRIGHT, J.E. 2001. New evidence for Cretaceous strike-slip faulting in the United States Cordillera and implications for terrane displacement, deformation patterns, and plutonism. American Journal of Science, 301, 150-181. YATES, R.G. 1968. The Trans-Idaho discontinuity. Proceedings of the 23rd International Geological Congress, 1968. Prague Academia, Prague, 117123.

This page intentionally left blank

The evolution of an exposed mid-lower crustal attachment zone in Fiordland, New Zealand KEITH A. KLEPEIS1 & GEOFFREY L. CLARKE2 l Department of Geology, University of Vermont, Burlington, VT 05405-0122, USA (e-mail: [email protected]) 2School of Geosciences, Division of Geology and Geophysics, University of Sydney, NSW 2006, Australia Abstract: Studies of convergent margins suggest that large subhorizontal shear zones in the lower crust help regulate how displacements are transferred horizontally and vertically through the lithosphere. We present structural data from the Fiordland belt of SW New Zealand that illustrate the progressive evolution of a 25 km thick section of exhumed, Early Cretaceous middle and lower crust. The data show that the mechanisms by which displacements were relayed through the crust during a 25 Ma cycle of arc-related magmatism, high-grade metamorphism and contraction changed repeatedly. During the period 126120 Ma, a >10km thick batholith composed of gabbroic-dioritic magma was emplaced into the lower crust. Melt-enhanced shear zones evolved at the upper and lower contacts of the batholith where magma and steep temperature gradients created strength contrasts. By ~120 Ma, partial melting of mafic-intermediate lower crust resulted in the formation of high-pressure (14-16kbar) migmatite and steep, regionally extensive vein networks up to 10 km below the batholith. Melt segregation and transfer through and out of the lower crust were aided by melt-induced fracture arrays and ductile deformation in shear zones. During the period 116-105 Ma, differential shortening of the crust produced a network of subhorizontal and subvertical shear zones at different crustal depths. Near-vertical shear zones up to 15 km wide formed at the deepest part of the section. These shear zones cut upwards across the entire lower crust to merge with a gently dipping upper amphibolite facies fold-and-thrust zone that formed in the middle crust. A 1 km thick, subhorizontal shear zone underlies this mid-crustal fold-and-thrust zone and physically connected shear zones that formed at different crustal depths. Our data suggest that deformation above and below this mid-lower crustal attachment zone was coupled kinematically and accommodated subhorizontal arc-normal displacements in the middle crust and oblique sinistral displacements on steep shear zones in the lower crust. The steep lower crustal shear zones also record components of subhorizontal arc-normal shortening and vertical thickening. These results strongly suggest that large, kinematically coupled networks of flat and steep shear zones separated the Fiordland crust into distinctive structural domains and relayed displacements vertically and horizontally through the lithosphere during Early Cretaceous oblique convergence.

The degree and mechanisms by which deformation in the upper crust is coupled to deformation in the lower crust and mantle are two of the least understood issues of continental dynamics. In convergent regimes, the middle and lower crust typically contain large flat or dipping shear zones that separate the lithosphere into distinctive structural domains (Fuis & Clowes 1993; Mayer et al 1991 \ Oldow et al 1990; Axen et al. 1998; Miller & Paterson 2001; Karlstrom & Williams 2002). Physical, analytical and numerical models of convergence (Harry et al. 1995; Roy den 1996; Teyssier et al. 2002; Ellis et al 1998; Beaumont et al 2001)

suggest that some of these shear zones act as transfer zones that help relay displacements vertically and horizontally through the lithosphere. However, exactly how these zones form and evolve through time are unresolved problems in lithospheric dynamics. The application of simple steady-state numerical models of lower crustal deformation to natural settings is complicated by the extremely heterogeneous and transient nature of lower crustal rheology and structure. Some studies (e.g. Karlstrom & Williams 2002; Miller & Paterson 2001) have shown that discrete, layered subdivisions at the deepest levels of the

From: GROCOTT, J., MCCAFFREY, K. J. W., TAYLOR, G. & TIKOFF, B. (eds) 2004. Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 197-229. 0305-8719/04/$15 © The Geological Society of London 2004.

198

K. A. KLEPEIS ETAL.

crust are unlikely to persist for very long, as the rheological profile of the crust responds to changes in temperature, fluid activity, magmatism and composition. This extreme variability implies that the manner in which mantle and lower crustal displacements are transferred through the lithosphere is also highly variable and reflects changing ambient conditions. Further complicating the picture, our understanding of the kinematic and mechanical behaviour of the lower crust for different scenarios has been hindered by a lack of natural examples. Inadequate exposure of the lower crust and the inherent difficulty in studying deformation evolving simultaneously within large sections of ancient lower crust have inhibited our understanding of lithospheric-scale processes. In addition, the insensitivity of most geophysical imaging techniques to the age and kinematic significance of lower crustal structures complicates the interpretation of deformation at the lithospheric scale. In this chapter, we describe the structural evolution of a 25 km thick, exhumed section of mafic lower crust in Fiordland, New Zealand, that formed at the roots of an Early Cretaceous magmatic arc (Fig. 1). Fiordland contains the Earth's largest belt (>5000 km2) of exhumed Mesozoic lower crustal granulites. These exposures not only provide us with an important natural example of the variability of lower crustal strength profiles, but also allow us to examine directly the effects of this variability on the evolution of lower crustal fabrics and shear zones. Through work related to this and our previous studies (Klepeis et al 1999, 2001, 2003; Clarke et al 2000; Daczko et al 2001<s, b, 2002), we identified a continuous lower crustal section in Fiordland representing Early Cretaceous paltoodepths of 25-50 km. This exposure, excellent geochronological control and regional-scale crosscutting relationships enabled us to determine how the lower crustal section evolved structurally within a well-constrained, ~25 Ma period of time (130-105 Ma).

Geological setting The geology of SW New Zealand records a history of magmatism, high-grade metamorphism and convergence that accompanied the evolution of an Early Mesozoic magmatic arc along the ancient margin of Gondwana (J.D. Bradshaw 1989; Muir et al 1994; Daczko et al 20010). Part of this arc is thought to have formed initially outboard of the Gondwana margin during Early Triassic-Early Cretaceous (247-131 Ma) sub-

duction-related magmatism (Kimbrough et al 1994; Muir et al 1998; Tulloch & Kimbrough 2003). In Fiordland (Fig. Ib), a north- and NNEstriking belt of plutonic, volcanic and sedimentary rocks called the Median Tectonic Zone (Tulloch & Kimbrough 1989; Muir et al 1994) or the Median Batholith (Mortimer et al 1999) represents this outboard part of the arc. On the eastern side of the Median Tectonic Zone (MTZ), Triassic plutons intruded Permo-Triassic volcanoclastic rocks of the Brook Street volcanics and the Maitai terrane (Fig. 2). These latter units represent arc-derived sedimentary rocks and accretionary complexes that also lay outboard of Gondwana during the Early Mesozoic (J.D. Bradshaw 1989; Mortimer et al 1999). Following its initial construction, the outboard part of the Mesozoic arc (the MTZ) accreted onto the Gondwana margin during convergence. This stage of arc evolution resulted in the emplacement of 126-105 Ma intrusive rocks (ages from Kimbrough et al 1984; Mattinson et al 1986; McCulloch et al 1987; J.Y. Bradshaw 1989; Tulloch & Kimbrough 2003; Klepeis et al 2004) into crust composed of both MTZ and Gondwana margin rocks (Mortimer et al 1999). Mafic-felsic rocks of the Western Fiordland Orthogneiss and the Separation Point Suite are included in this event (Fig. Ib). In addition, the ages of mafic dykes that intrude Lower Paleaozoic host gneiss in the Arthur River complex (Fig. 2) suggest that the amalgamation of the outboard arc with Gondwana may have occurred as early as ~ 136 Ma and probably by -129 Ma (Hollis et al 2002). In Fiordland, rocks of the Gondwana margin are represented by metasedimentary rock and granitic orthogneiss of Cambrian-Permian age (Fig. Ib). Intense contractional deformation and crustal thickening accompanied and followed magmatism that resulted in the emplacement of the Western Piordland Orthogneiss (Bradshaw 1990; Muir etal 1995, 1998; Daczko etal 2001(3, 2002). On the basis of geochronological and geochemical data, Muir et al (1995, 1998) suggested that the sudden appearance of large volumes of Na-rich magma within the arc at —126 Ma, including the Western Fiordland Orthogneiss, was triggered tectonically by the underthrusting and subsequent melting of MTZ rocks beneath western Fiordland. Daczko et al (2001a, 2002) presented structural data that support this interpretation. However, the origin of this contraction is controversial. The deformation may reflect the collision of the outboard part of the arc (the MTZ) with Gondwana following emplacement of the Western Fiordland Orthogneiss (Bradshaw 1990; Bradshaw &

MID-LOWER CRUSTAL ATTACHMENT ZONE

199

Fig. 1. (a) Location of Fiordland on the South Island of New Zealand, (b) Geological map of Fiordland showing regional subdivisions. Pressures representing the peak of Early Cretaceous granulite facies metamorphism at ~120 Ma show tilted lower crustal section constructed using metamorphic data in Table 1 and, for the Doubtful Sound area, in Gibson & Ireland (1995). (c) Schematic cross-section showing a convergent setting between the Gondwana continental margin and an Early Mesozoic arc (MTZ). WFO, Western Fiordland Orthogneiss batholith.

Kimbrough 1989; Daczko et al 2001^, 2002). Alternatively, the deformation could reflect the collision of a basaltic plateau with the subduction zone on the outboard side of the arc (Sutherland & Hollis 2001; Tulloch & Kimbrough 2003). In addition to the deformation, crustal thickening is reflected in the occurence of high-pressure granulite facies and upper amphibolite facies metamorphism that recrystallized the Western Fiordland Orthogneiss and its host rock (Blattner 1978; J.Y. Bradshaw 1989; Clarke et al 2000;

Daczko et al. 2001/?). Garnet-pyroxeneplagioclase-bearing assemblages have yielded peak metamorphic pressures of P = 1216kbar and temperatures of r > 7 5 0 ° C (Table 1). These assemblages occur within the Western Fiordland Orthogneiss and its host rocks between Doubtful Sound and Milford Sound (Fig. 1). The Early Cretaceous age of this metamorphism is constrained by three relationships: (1) crystallization ages (126120 Ma) of the Western Fiordland Orthogneiss

200

K. A. KLEPEIS ETAL

Fig. 2. Geological map of north-central Fiordland showing the major lithological subdivisions. Map was constructed using data from this study and from Bradshaw (1989, 1990), Blattner (1991), Turnbull (2000), Daczko et al. (2002) and Claypool (2002). Plotted dates (see also Table 2) are from U-Pb analyses of zircon cores and rims reported by Hollis et al. (2002). Data show the distribution of Early Cretaceous crystallization (zircon core) ages and ~120 Ma metamorphic (rim) ages from the Arthur River complex. Also note distribution of migmatite (m) below the batholith. MD, Mount Daniel; ME, Mount Edgar; P, Pembroke Valley; CO, Camp Oven Creek; ARC, Arthur River complex; MTZ, Median Tectonic Zone. Profiles A-A', B-B', C-C, D-D', E-E' and F-F are shown in Fig. 3.

(Table 2); (2) metamorphic ages from the Arthur River complex (Fig. 2) and adjacent units; and (3) crosscutting relationships between highgrade fabrics and dykes of known age (Mattinson et al. 1986; McCulloch et al. 1987; Gibson & Ireland 1995; Tulloch et al. 2000; Hollis et al. 2002; Klepeis et al. 2004). The ages of specific fabrics and rock units discussed in this chapter are presented in more detail below. By ~ 105 Ma, widespread extension affected parts of the Fiordland belt and adjacent areas (ID. Bradshaw 1989; Tulloch & Kimbrough

1989; Gibson & Ireland 1995). The Doubtful Sound shear zone (Fig. Ib) is the dominant structural expression of this extension in Fiordland. This shear zone is composed of granulite facies and upper amphibolite facies fabrics that cut the Western Fiordland Orthogneiss. Kinematic data reported by Gibson et al. (1988) and Claypool (2002) indicate that this shear zone records NE-SW stretching and a dominantly NE-directed sense of shear that is consistent with extension directions in other parts of New Zealand during the Mid-Cretaceous (Tulloch & Kimbrough

Table 1. P-T data from northern Fiordland between Caswell and Milford Sounds Location

Lithological unit*

Assemblage

Calculated P (kbar)

Calculated T (°C)

P-T method*

Data source

Interpretation

g-bi-hbl-pl-q

4-5

500-600

6,8,9

Daczko et al (2002)

Palaeozoic tectonism

Pembroke Valley

Country rock outside WFO contact aureole

ARC

opx-cpx-hbl

<8

>750

see source

Early stages of

George Sound

WFO contact aureole Country rock rafts in WFO

g-bi-hbl-pl

8.7 ± 2.1

470, 482, 604,

6,8,9

g-bi-pl-ky-q

11-12

704-735

see source

g-bi-ksp-pl

7-8

>700

6,8,9

Charles Sound

WFO and contact aureole

WFO

g-cpx-pl-q

8-10

Mount Daniel & adjacent areas

WFO and ARC

g-cpx-pl-ksp-rt-q

12-13

650-700

see source

Milford Sound

ARC

to

2-cpx-pl-q-kv r r M J

13-15

730-870

1, 2, 3, 4, 5, 7

Milford Sound

ARC

g-cpx-pl-q

14, 16

715-812

1, 2, 3, 4, 5

Milford Sound

ARC

g-cpx-hbl-pl-q

13-15

800

1, 2, 3, 4, 5

Mount Daniel Milford Sound

ARC and WFO ARC

g-cpx-ky-pl-hbl-q g-cpx-pl-q

13.9 11.8-13.7

700-800 624-775

1,2,5 1, 2, 3, 4, 5

Poison Bay

ARC

g-cpx-hbl-pl-q

12.4-15.6

768-820

1, 2, 3, 4, 5

Pembroke Valley

ARC

w g-cpx-pl-q to .f -^ F i

14.7-16.6

662-700

1, 2, 3, 4, 5, 6

Mount Daniel & adjacent areas Poison Bay

ARC and WFO

g-cpx-bi-pl

11-12

662-766

see source

WFO

g-cpx-pl-q

12.6 ± 1.9

718+ 115

8

Clarke et al (2000) Daczko et al. (2002) Bradshaw (1985, 1989) Daczko et al. (2002) Bradshaw (1985) Bradshaw (1985, 1989) Clarke et al. (2000) Clarke et al. (2000) Clarke et al. (2000) Daczko (2001) Clarke et al. (2000) Clarke et al. (2000) Clarke et al. (2000) Bradshaw (1989) Clarke et al. (2000)

Caswell Sound

George Sound & Mount Daniel areas Caswell Sound

686

see source

arc

Stages 1 and 2 Stages 1 and 2 Stages 2 and 3 Stage 2 Stage 2 Stage 2 Stage 2 Stage 2 Stage 2 Stage 2 Stage 2 Stage 2 Stage 2 Stage 2 (continued}

Table 1. Continued Location

Lithological unit*

Assemblage1

Calculated P (kbar)

Milford Sound

ARC

g-hbl-pl-q-ky

13-16

Milford Sound

ARC

15.8 ± 2.6

Pembroke Valley

ARC

Pembroke Valley

ARC

g-hbl-pl-bi-epky-q g-cpx-pl-rt-hblcz-q g-bi-pl-rt-hbl-cz-q

Pembroke Valley & Mount Daniel Pembroke Valley & Mount Daniel Pembroke Valley & Mount Daniel Poison Bay

ARC

Poison Bay

Calculated T (°Q 674

P-T method* 6, 7

Data source

Interpretation

14.0 ± 1.3

676 + 34

1, 3, 4, 6, 8, 11

14.1 ± 1.2

674 + 36

1, 3,4, 6, 8, 11

g-hbl-cz-pl-q-ky

11.7-13.3

677 + 64

6,7, 8, 11

Clarke et al. (2000) Clarke et al (2000) Daczko et al (2001«) Daczko et al. (20010) Daczko (2001)

ARC

g-hbl-pa-pl-cz-q

11.4, 11.1, 13.2 ± 1.7

687-694, 703-736

6, 8, 11

Daczko (2001)

Stage 3, cooling

ARC

g-hbl-mu-pl-cz-q

10.9, 11.9+ 1.8

704, 629 ± 56

6, 8, 11

Daczko (2001)

Stage 3, cooling

ARC

g-pl-bi-hbl-rt-cz

11.9+ 1.1

581 + 34

8

Exhumation

ARC

g-hbl-pl-bi-ti-cz

8.7 + 1.2

587 + 42

8

Klepeis et al. (1999) Klepeis et al. (1999)

8

Stage 3 Stage 3 Stage 3, duplex Stage 3, duplex Stage 3, cooling

Exhumation

*ARC, Arthur River complex; WFO, Western Fiordland Orthogneiss; results organized according to stages discussed in the text (oldest to youngest). f g, garnet; cpx, clinopyroxene; opx, orthopyroxene; pi, plagioclase; q, quartz; ky, kyanite; bi, biotite; ep, epidote; hbl, hornblende; rt, rutile, cz, clinozoisite; ksp, potassium feldspar; ti, titanite; pa, paragonite; mu, muscovite. *1, Ellis & Green (1979); 2, Powell (1985); 3, Krogh (1988); 4, Eckert et al. (1991); 5, Newton & Perkins (1982); 6, Graham & Powell (1984); 7, Newton & Haselton (1981); 8, Powell & Holland (1988); 9, Ferry & Spear (1978); 10, Hodges & Spear (1982); 11, Kohn & Spear (1990).

Table 2. Geochronological data from intrusive rocks in northern Fiordland* Lithological unit1

Rock type*

Method*

Mineral

Corrected age (Ma)

Dyke in Milford Gneiss Milford Gneiss

pegmatite

U-Pb ion probe

zircon

81.8 ± 1.8

metadiorite

K-Ar

hornblende

90.3 ± 2.6

Milford Gneiss

metadiorite

K-Ar

hornblende

91 ± 0.4

Milford Gneiss

diorite gneiss

K-Ar

hornblende

92.3 ± 2.8

WFO

metadiorite

U-Pb SID

apatite

91 ±2

WFO

amphibolite

K-Ar

hornblende

92.6 ± 0.9

WFO

metadiorite

U-Pb SID

apatite

92.6 ± 2

Milford Gneiss

hbl gneiss

K-Ar

hornblende

115-74

Milford Gneiss Milford Gneiss

hbl gneiss hbl gneiss

K-Ar K-Ar

hornblende hornblende

105 105 ± 4

Milford Gneiss

gneiss

K-Ar

hornblende

106.4 ± 0.82

WFO

diorite gneiss

U-Pb ion probe

zircon

107.5 ± 2.8

Milford Gneiss

hbl granulite

K-Ar

hornblende

111 ± 2

Milford Gneiss

dioritic granulite

K-Ar

plagioclase

113.4 ± 1.8

Darran complex

gabbronorite

U-Pb SID

apatite

113 ±0.4

MTZ

leucogranite

K-Ar

biotite

114.8 + 5.1

Interpretation"

Source Hoflis et al (2002) Nathan et al. (2000) Nathan et al. (2000) Nathan et al. (2000) Mattinson et al. (1986) Gibson et al. (1988) Mattinson et al. (1986) Blattner(1991) Blattner (1978) Nathan et al. (2000) Nathan et al. (2000) Gibson & Ireland (1995) Nathan et al. (2000) Nathan et al (2000) Mattinson et al. (1986) Williams & Harper (1978)

C or inherited

Exhumation phase

Cooling

Exhumation phase

Cooling

Exhumation phase

Cooling

Exhumation phase Exhumation phase

Cooling

Exhumation phase Exhumation phase

M/cooling M M

End Stage 3 exhumation End Stage 3 End Stage 3

M

End Stage 3

M

Stage 3

M

Stage 3

M

Stage 3 Stage 3

C

Stage 3

(continued}

Table 2. Continued Lithological unitf

Rock type*

Method§

Mineral

Corrected age (Ma)

MTZ

trondhjemite

K-Ar

biotite

115.8 ± 3.1

MTZ

trondhjemite

K-Ar

muscovite

115.8 ±4.1

Harrison Gneiss

orthogneiss

U-Pb ion probe

zircon rim

120 ± 2

Palaeozoic gneiss in ARC1 Selwyn Creek Gneiss11 Dyke in Milford Gneiss1 WFO

migmatitic gneiss

U-Pb ion probe

zircon rim

120

dioritic orthogneiss felsic dyke (02DA)

U-Pb ion probe

zircon rim

120

U-Pb ion probe

zircon rim

123

amphibolite

U-Pb ion probe

zircon

119 ± 5

WFO

orthogneiss

U-Pb SID

zircon

120-130

WFO

granulite

Rb-Sr

whole rock

120 ± 15

WFO

orthognesis

U-Pb ion probe

zircon

125.9 ± 1.9

WFO

granulitic metabasite

U-Pb ion probe

zircon

126 + 3

Harrison Gneiss

orthogneiss

SHRIMP

zircon

134 + 2

Pembroke Granulite Mount Edgar Diorite1

mafic granulite

K-Ar

hornblende

130

dioritic orthogneiss

U-Pb ion probe

zircon

128.8 ± 2.4

Source Williams & Harper (1978) Williams & Harper (1978) Tulloch et al (2000) Hollis et al (2002) Hollis et al (2002) Hollis et al (2002) Gibson & Ireland (1995) Mattinson et al (1986) McCulloch etal (1987) Muir et al (1998) Gibson & Ireland (1995) Tulloch et al (2000) Blattner(1991) Hollis et al (2002)

Interpretation" C

Stage 3

C

Stage 3

M

Stages 1 and 2

M

Stages 1 and 2

M

Stages 1 and 2

M

Stages 1 and 2

C

Stages 1 and 2

C

Stages 1 and 2

C

Stages 1 and 2

C

Stages 1 and 2

C

Stages 1 and 2

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

Dyke in Milford Gneiss11 Milford Gneiss1

felsic dyke (02DA)

U-Pb ion probe

zircon core

133.4 ± 2.1

U-Pb ion probe

zircon

131.3 ± 2.9

U-Pb ion probe

zircon

132.6 ± 2.9

Dyke in Milford Gneiss1 Milford Gneiss

gabbroic orthogneiss dioritic orthogneiss diorite dyke

U-Pb ion probe

zircon

135.9 ± 1.8

granulite

K-Ar

hornblende

140.1 ± 1.02

Milford Gneiss

granulite

K-Ar

hornblende

138 ± 4

Milford Gneiss

hbl schist

K-Ar

hornblende

149 + 4

Milford Gneiss

amphibolite

K-Ar

hornblende

164 ± 4

Selwyn Creek Gneiss1 Darran complex

dioritic orthogneiss diorite

U-Pb ion probe

zircon core

154.4 ± 3.6

K-Ar

biotite

134 + 4

Darran complex

gabbronorite

SID

apatite

135 ± 0.4

Darran complex

diorite

K-Ar

biotite

135 + 4

Darran complex

U-Pb SID

apatite

135.6 ± 0.4

Darran complex

olivine gabbronorite felsic dyke

U-Pb ion probe

zircon

136.8 + 1.9

Darran complex

gabbronorite

U-Pb SID

zircon

137 ±4

Darran complex

gabbronorite

U-Pb SID

zircon

137 ± 1

Darran complex

diorite

U-Pb ion probe

zircon

138 ± 2.9

Milford Gneiss11

Hollis et al (2002) Hollis et al (2002) Hollis et al. (2002) Hollis et al (2002) Nathan et al (2000) Nathan et al (2000) Nathan et al (2000) Nathan et al (2000) Hollis et al (2000) Nathan et al (2000) Mattinson et al (1986) Nathan et al (2000) Mattinson et al (1986) Muir et al (1998) Kimbrough et al (1994) Mattinson et al (1986) Muir et al. (1998)

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Stages 1 and 2

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc (continued}

Table 2. Continueaf Lithological unit1

Rock type*

Met!i0df

Mineral

Corrected age (Ma)

Darran complex

leucogabbro

K-Ar

biotite

140 ±2

Darran complex

monzite

U-Pb SID

zircon

141 ± 4

Darran complex

diorite

U-Pb SID

zircon

142 ± 5

Darran complex

diorite

K-Ar

hornblende

142 ±4

Darran complex

dioritic pegmatite

K-Ar

hornblende

142 + 6

Darran complex Darran complex

dioritic pegmatite diorite

K-Ar K-Ar

hornblende biotite

142.8 + 1.8 143.82 ± 1.2

Darran complex

dioritic pegmatite

K-Ar

hornblende

145.2 ± 3

Darran complex

diorite

K-Ar

biotite

149 ± 2

Darran complex

diorite pegmatite

K-Ar

hornblende

153 ± 8

Darran complex

hbl dioritic pegmatite hbl dioritic pegmatite

K-Ar

hornblende

152.6 ± 2.1

K-Ar

hornblende

156.4 + 4.1

Darran complex

Source Nathan et al. (2000) Kimbrough et al. (1994) Kimbrough etal. (1994) Nathan et al (2000) Nathan et al. (2000) Adams (1975) Nathan et al. (2000) Nathan et al. (2000) Nathan et al. (2000) Nathan et al. (2000) Adams (1975) Nathan et al. (2000)

*Data are organized according to stages discussed in the text (from youngest to oldest) Additional unpublished ages cited in the text. f ARC, Arthur River complex; WFO, Western Fiordland Orthogneiss; MTZ, Median Tectonic Zone. *hbl, hornblende, § SID, standard isotope dilution method. "C, crystallization age; M, metamorphic age. " Sample location indicated on Figure 2.

Interpretation" C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C C

Early phase of arc Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

C

Early phase of arc

MID-LOWER CRUSTAL ATTACHMENT ZONE

1989). The crystallization age of the Western Fiordland Orthogneiss and K-Ar cooling ages on hornblende from the upper amphibolite facies fabrics indicate a Mid-Cretaceous (~ 108 Ma) age for this deformation (Gibson et al 1988). The Anita shear zone (Figs Ib & 2) in northern Fiordland also formed during or after this period (Hill 1995; Klepeis et al 1999). Both of the Doubtful Sound and Anita shear zones record cooling, decompression and exhumation of the granulite belt following crustal thickening and arc-related magmatism (Gibson & Ireland 1995; Klepeis et al 1999). Crustal structure and geochronology In this section we define the boundaries and main lithological divisions of Fiordland's highpressure metamorphic belt (7-16kbar, Fig. Ib, Table 1). These rocks represent the deformed and metamorphosed roots of the Early Mesozoic arc. We also review the ages of the major rock units that make up the section (Fig. 2, Table 2) and describe structural relationships (Figs 3 & 4) that reflect a heterogeneous history of Early Cretaceous magmatism and convergence. Boundaries of the high-grade metamorphic belt Rocks of contrasting tectonic affinity surround Fiordland's high-pressure granulite and upper amphibolite facies gneisses. To the east of the high-grade rocks, the Darran Suite is composed of mostly unmetamorphosed gabbro and diorite that represent the upper crustal part of the Early Mesozoic arc (the MTZ). Most of the Darran Suite is only weakly deformed. However, its western margin is highly deformed and shares a steep, upper amphibolite facies foliation that forms the dominant structural grain north and NE of Milford Sound (Figs 3b & 4). This foliation represents part of a steep, 10-15 km wide shear zone (Figs 3b & 4) defined and discussed in detail later in this chapter. SW of the Darran Suite, weakly foliated tonalite and quartz diorite of the Roxburgh Suite (Fig. 2) form the boundary between the high-grade gneisses to the west and low-grade plutonic rocks and Tertiary sedimentary rocks of the MTZ to the east. The Roxburgh Suite is thought to be mostly Palaeozoic in age (Turnbull 2000). The western boundary of the high-grade gneisses coincides with the 4-5 km wide, nearvertical Anita shear zone (Fig. 2). The Anita shear zone separates intensely deformed granulite and upper amphibolite facies rocks of the

207

Arthur River complex from weakly deformed Lower Palaeozoic metasedimentary rocks and granite of the Gondwana margin. The Greenland Group, the Thurso Gneiss and the Saint Anne Gneiss (Fig. 2) represent parts of this Lower Palaeozoic succession. Upper amphibolite and greenschist facies mineral assemblages within the Anita shear zone record peak metamorphic pressures of — 12kbar and decompression that accompanied exhumation of the Fiordland granulites after -108-105 Ma (Hill 1995; Klepeis et al 1999). The Alpine fault and other Late Cenozoic faults (Fig. 2) reactivate the steep margins of the Anita shear zone and truncate the exposures of high-grade gneisses north of Milford Sound (Claypool et al 2002). The southern boundary of the metamorphic belt occurs south of Doubtful Sound (Fig. Ib) and coincides with a large Tertiary fault that forms the northern margin of the Cambrian-Permian Southwest Fiordland terrane (Fig. Ib). Lithological divisions of the magmatic arc West of the Darran Suite near Milford Sound, the Arthur River complex (ARC, Fig. 2) is composed of gabbroic and dioritic gneiss that is subdivided into four lithologically distinctive units (Fig. 2). The Milford Gneiss is composed mostly of garnet- and hornblende-bearing metagabbro with minor migmatitic metadiorite (m, Fig. 2). Mafic dykes of the Milford Gneiss intrude rafts of Palaeozoic rock at Camp Oven Creek south of Milford Sound (CO, Fig. 2). The Mount Edgar Diorite (Fig. 2) is a pile of dioritic sheets that intruded into the Milford Gneiss, cutting across older gneissic layering at low angles. Structural relationships within these units are described in more detail in the next section. North of Milford Sound, the Pembroke Granulite forms a large pod that is aligned parallel to the NNE strike of the dominant foliation near Milford Sound (Fig. 4). This unit includes two pyroxene- and hornblende-bearing metagabbros and migmatitic metadiorite. East of the Pembroke Granulite, the Harrison Gneiss is composed of banded metadiorite with discontinuous sheets of mafic to felsic compositions in gradational contact with the more mafic Milford Gneiss. The zone located between the Arthur River complex and the Darran Suite is composed of discontinuous gabbroic units intruded by variably deformed quartz diorite gneiss and felsic dykes. Bradshaw (1990) divided this zone into the Indecision Creek and Mount Anau igneous complexes. For simplicity, we refer to all of the rocks in this area as the Indecision Creek complex (Fig. 2). The northern contact of the

208

K. A. KLEPEIS ETAL.

Fig. 3. (a) Composite cross-section constructed using structural data from Caswell, Charles and George Sounds (locations shown in Fig. 2). Profile shows the geometry of structures above and below the upper contact of the Western Fiordland Orthogneiss batholith. Shaded areas represent Palaeozoic paragneiss representative of the ancient Gondwana margin. Dashed lines near C-C represent gneissic layering that predated batholith emplacement. (b) Composite cross-section constructed using data from the region between the Pembroke Valley and Mount Daniel (for locations see Fig. 2). Sections show the geometry of structures above and below the lowermost contact of the Western Fiordland Orthogneiss. Patterns are the same as in Figure 2.

Indecision Creek complex is gradational with the Harrison Gneiss; its eastern and western contacts are gradational with dioritic rocks of the weakly deformed Darran Suite and the highly deformed Western Fiordland Orthogneiss, respectively. This unit is chemically similar to the Western Fiordland Orthogneiss (Fig. 2) and may contain less Palaeozoic inheritance than the subunits of the Arthur River complex (Turnbull 2000). In the central part of the high-grade metamorphic belt, the Western Fiordland Orthogneiss is a major batholith that intruded the Arthur

River complex. The batholith is composed mostly of layered sequences of gabbro and diorite with some ultramafic dykes and pods. The lower (northern) contact of this unit is well exposed at Mount Daniel (MD, Fig. 2); its upper (southern) contact with Palaeozoic metasedimentary rocks is well exposed at Caswell Sound. Some areas of this batholith preserve primary igneous layering that escaped the intense recrystallization observed in most of the rock units of northern Fiordland (described below). However, in areas such as the eastern end of George Sound and along its eastern

MID-LOWER CRUSTAL ATTACHMENT ZONE

209

Fig. 4. (a) Map showing structural data from northern Fiordland. Structural data are from this study and data from Bradshaw (1989, 1990), Blattner (1991), Klepeis et al (1999), Daczko et al (2002) and Claypool et al (2002). Foliation trajectories represent the interpolation of structural trends constructed using plotted data and reconnaissance mapping. Shaded areas show the Indecision Creek shear zone (eastern side) and the George Sound shear zone (western side), Lower hemisphere. Equal-area stereoplots in (b), (c), (d) and (e) show poles to foliations (open circles) and mineral lineations (black squares) for the four structural domains discussed in the text. Black lines in stereoplots represent best-fit planes to foliation clusters.

210

K. A. KLEPEIS ETAL.

contact, these rocks were intensely deformed and recrystallized. Rafts of Palaeozoic metasedimentary rocks occur within and near the contacts of the intrusion east of Mount Daniel and at George Sound. These metasedimentary rocks are migmatitic (m, Fig. 2) near the contacts with the batholith.

Ages of magmatism, crustal melting and high-grade metamorphism Published U-Pb ages from northern Fiordland (Fig. 2, Table 2) reflect a history of Early Cretaceous magmatism and Early Cretaceous metamorphism and melting. Tulloch et al (2000) analysed zircon cores from the Arthur River complex near Milford Sound and obtained both Early Carboniferous (—360 Ma) and Early Cretaceous (— 134 Ma) U-Pb ages. Hollis et al (2002) obtained 136-129 Ma U-Pb ages from the cores of magmatic zircon in the Arthur River complex and showed that these ages most likely reflect the crystallization of Early Cretaceous dykes and plutons. These data, combined with evidence of an intrusive relationship between the mafic dykes of the Milford Gneiss and migmatitic Palaeozoic host rocks at Camp Oven Creek (CO, Fig. 2), suggest that the Arthur River complex was emplaced into crust composed of a mixture of MTZ arc rocks and Palaeozoic rocks of Gondwana. A crystallization age of — 154 Ma for the main dioritic part of the Selwyn Creek Gneiss (Fig. 2, Table 2) suggests that these rocks also were emplaced during an early phase of MTZ magmatism (Hollis et al. 2002) and likely represent deformed parts of the Darran Suite. Following emplacement of the Arthur River complex, the Mount Edgar Diorite (Fig. 2) was emplaced at - 129 Ma (Hollis et al 2002). This age is consistent with field relationships indicating that the Mount Edgar Diorite intruded and cuts across gneissic layering in the Milford Gneiss at low angles. The dioritic parts of the Indecision Creek complex also intruded older metagabbroic crust. A few unpublished U-Pb dates suggest that parts of it intruded between — 135 and — 122 Ma (reported in Bradshaw 1985, 1990). These ages, combined with a history of some Palaeozoic inheritance (although less than the Arthur River complex), indicate that it was emplaced following the amalgamation of the MTZ with the Gondwana margin. This relationship suggests that it probably has a similar origin as the Mount Edgar Diorite and the Western Fiordland Orthogneiss.

U-Pb geochronological data also indicate that high-grade metamorphism accompanied Early Cretaceous magmatism and crustal melting. Thin, low-U metamorphic rims around magmatic zircon cores of both Palaeozoic and Early Cretaceous occur in the Milford Gneiss, the Harrison Gneiss, the Selwyn Creek Gneiss and Palaeozoic gneiss at Camp Oven Creek (Tulloch et al 2000; Hollis et al 2002; Klepeis et al 2004). These ages are interpreted to reflect the influx of heat that accompanied the emplacement of the mafic-intermediate Western Fiordland Orthogneiss (Tulloch et al 2000; Hollis et al 2002). This interpretation is consistent with the range of crystallization ages from the Western Fiordland Orthogneiss (126-120 Ma, Table 2) and with the spatial distribution of zircon displaying metamorphic rims in migmatite and granulite facies rocks below the batholith. The age of lower crustal melting (migmatite localities shown in Fig. 2) also is interpreted to have coincided with this widespread thermal pulse. Zircon ages (81.8 ± 1.8 Ma) from a posttectonic dyke located in the Pembroke Valley (P, Fig. 2) north of Milford Sound indicate that high-grade granulite and upper amphibolite facies metamorphism in the Arthur River complex terminated in the Mid-Cretaceous (Hollis et al 2002). In addition, K-Ar ages on hornblende (Gibson et al 1988; Nathan et al 2000) and U-Pb dates on apatite (Mattinson et al 1986) indicate that the Western Fiordland Orthogneiss and Arthur River complex had cooled to 300-400C by -90 Ma (Table 2). KAr amphibole and biotite cooling ages of —93 and —77 Ma, respectively, also support a MidCretaceous age for the extensional Doubtful Sound shear zone (Gibson et al 1988). Structural relationships Two distinctive structural domains characterize the zone between the Anita shear zone and the Darran Suite (Fig. 4a-c). The western part of this zone is dominated by layered intrusions (e.g. the Mount Edgar Diorite) and gneissic foliations that mostly dip moderately to the south, SW and west (Fig. 4b). Hornblende and biotite mineral lineations on foliation planes plunge moderately to the west and SW. This western domain is narrowest (>10km) near Milford Sound and widest (>30km) near Caswell Sound. The eastern domain contains penetrative upper amphibolite facies foliations that are mostly steep to subvertical and strike to the NNE (Figs 3b & 4c). These latter foliations

MID-LOWER CRUSTAL ATTACHMENT ZONE

define a 10-15 km wide zone of penetrative ductile deformation that we define in this chapter as a steep shear zone. Hornblende and biotite mineral lineations in this domain mostly plunge moderately to the south and SW with some localities also displaying near-vertical plunges (Fig. 4c). We introduce the new name Indecision Creek shear zone for this zone of intense deformation. All foliations and lithological contacts in both domains are cut by mylonitic foliations of the Anita shear zone (Fig. 4a). The structure of the western domain constitutes a tilted middle and lower crustal section, with the deepest palaeodepths occurring in the north near Milford Sound and the shallowest palaeodepths occurring in the south near Caswell and George Sounds (Fig. Ib, Table 1). The dominant features in this section include the diorite intrusions that make up the Mount Edgar Diorite and the Western Fiordland Orthogneiss (Figs 2 & 3c). At Mount Daniel (MD, Fig. 2) the lower contact of the Western Fiordland Orthogneiss is mostly concordant with gneissic foliations in the Milford Gneiss. This relationship suggests that magma exploited gneissic layering in the older rocks during batholith emplacement. Elsewhere, such as at the northwestern end of George Sound, the contacts of the Western Fiordland Orthogneiss cut obliquely across gneissic layering in host rocks (Fig. 3a). In some areas, rocks of the Western Fiordland Orthogneiss (WFO) and Mount Edgar Diorite record a heterogeneous, tectonic overprint. The eastern side of the Mount Edgar Diorite is folded (Fig. 2). At Mount Daniel the lower contact of the WFO dips to the west and SW and also is folded (Fig. 3b). However, most of the original intrusive relationships between different phases of the batholith are well preserved at this locality. In areas where this deformation is intense, such as at the SE end of George Sound and near the upper and eastern contacts of the Western Fiordland Orthogneiss, these rocks contain a strong upper amphibolite facies tectonic foliation that transposes all primary igneous layering in the batholith. Thermobarometric data (Table 1) derived from garnet granulite and upper amphibolite facies mineral assemblages from the Western Fiordland Orthogneiss and its contact aureoles constrain the range of palaeodepths represented in the tilted section. Pressures reflecting the peak of Early Cretaceous metamorphism range from 7-9kbar at Caswell and George Sounds, to 10-13kbar near the base of the Western Fiordland Orthogneiss, and 13-16 kbar in the deepest part of the section north of Milford Sound (Fig. la). These

211

data indicate palaeodepths of 25-50 km from south to north (Fig. 3a, b). The dominant NNE-striking foliations of the eastern domain everywhere cut and transpose the west- and SW-dipping foliations and igneous layering of the western domain. Crosscutting relationships between these two groups of foliations are best preserved inside a ~15 km wide transitional zone that includes the eastern contacts of the Milford Gneiss (from east of Mount Edgar, ME, to the Pembroke Valley, P, Fig. 2) and the Western Fiordland Orthogneiss. In this transitional zone the dominant SW- and west-dipping foliations of the western domain are tightly folded into a series of overturned south-plunging folds (Figs 3b & 4a). Fold tightness (interlimb angle) increases from NW to SE across this zone and fold axial planes progressively steepen toward the SE. Where these folds are tight to isoclinal and upright, such as occurs inside the Indecision Creek complex, the NNE-striking foliation of the eastern domain parallels the axial planes of the folds. In zones of the highest strain, the steep NNE-striking foliations transpose all folds and intrusive contacts, including the eastern contacts of the Western Fiordland Orthogneiss and Mount Edgar Diorite. Mylonitic foliations also occur locally. These crosscutting relationships and variations in the intensity of folding and transposition form the basis of our interpretation that this eastern domain represents a zone of high strain. The kinematic significance of this zone is discussed in a later section (Stage 3). Between Caswell Sound and Mount Daniel, a third structural domain occurs near the southern end of George Sound (Fig. 4a). This domain exhibits structural relationships that are similar to those of the eastern domain. At George Sound, steep upper amphibolite facies foliations that strike to the NNE transpose older west- and SW-dipping gneissic foliations and dykes (Fig. 4d) inside the Western Fiordland Orthogneiss. These steep foliations parallel the axial planes of tight, upright folds of the dykes and the older gneiss. The boundaries of this zone approximately parallel those of the eastern domain (Fig. 4a). These relationships indicate that the structures of this third area, like those of the eastern domain, are younger than the dominant gneissic foliations of the titled section in western Fiordland. The tightness of folding and degree of transposition also indicate that this area represents a zone of high strain. We introduce the name George Sound shear zone to describe this zone of high strain. Finally, the fourth structural domain occurs at Caswell Sound (Fig. 4a, e). Here, a series of

212

K. A. KLEPEIS ETAL.

upper amphibolite and garnet granulite facies thrust zones cut and transpose the upper contact of the Western Fiordland Orthogneiss. The zone consists of both west- and east-verging ductile thrusts that separate a central domain that contains open to tight folds of gneissic foliation in Palaeozoic metasedimentary rocks (Fig. 3a). Fold tightness increases towards the contacts of the Western Fiordland Orthogneiss. The most intense zone of thrusts occurs at the eastern end of the sound. Here, thrust splays and fold axial surfaces curve downward to sole into a 1 km thick basal shear zone that occurs within the Western Fiordland Orthogneiss (Fig. 4e). Hornblende, clinozoisite and biotite mineral lineations on foliation surfaces in the dipping thrust zones plunge moderately and gently to the west (Fig. 4e). This basal shear zone is exposed at Charles Sound and at the eastern end of Caswell Sound. The steep upper amphibolite facies fabric of the George Sound shear zone merges into parallelism with this gently dipping basal shear zone (Figs 3a & 4a).

Space-time correlation of high-grade fabrics Crosscutting relationships, published U-Pb ages and similarities in metamorphic grade and structural style allowed us to correlate magmatic, metamorphic and deformational events across northern Fiordland. These correlations helped us to reconstruct the Cretaceous structural and magmatic evolution of the middle and lower crustal section. A key marker horizon in the section is the 126-120 Ma Western Fiordland Orthogneiss. The well-documented age and regional extent of this unit allowed us to divide the evolution of rock fabrics in the section into three time intervals corresponding to events that occurred before (>126Ma), during (126-120 Ma) and after (<120Ma) its emplacement. Rock fabrics that predate emplacement of the Western Fiordland Orthogneiss occur mostly in gneiss of the Arthur River complex and in Palaeozoic host rock near the margins of the batholith. Rock fabrics that formed during emplacement of the batholith occur within 1-2 km of the batholith contacts (e.g. at Mount Daniel and in the Palaeozoic rafts of George Sound, Fig. 2). Rock fabrics in these latter areas display textures that indicate that deformation occurred while the batholith was still partially molten. We describe these features in a later section. Another key relationship involves the Anita shear zone, which cuts all structures of the high-

grade gneiss belt at its northern and western ends (Fig. 4a). Because the Anita shear zone records decompression that began during regional extension at ~ 108-105 Ma, the Indecision Creek and George Sound shear zones must have evolved before this ~ 108-105 Ma interval. Finally, deformation in both the Indecision Creek shear zone and George Sound shear zone transposed all fabrics that formed during or before emplacement of the Western Fiordland Orthogneiss (Figs 3b & 4a). The subsolidus style and upper amphibolite facies mineral assemblages that characterize these shear zones indicate that they must have evolved after the Western Fiordland Orthogneiss had cooled and crystallized. Klepeis et al. 2004 report age of —116 Ma for deformation in the Indecision Creek shear zone. To highlight the sequential evolution of the different structural zones, we grouped all foliations in the high-grade gneisses and plotted them (Fig. 4a) according to their relative age with respect to the dominant foliations of the Indecision Creek shear zone. On the basis of similar orientations of structures, similar metamorphic grade and identical crosscutting relationships with respect to the Western Fiordland Orthogneiss and the Anita shear zone, we correlated the steep foliations of the Indecision Creek and George Sound shear zones. Foliations that formed during the deformation that produced the shear zones are plotted with white triangles; those that predate the shear zone are plotted with black triangles (Fig. 4a). The mylonitic foliations of the Anita shear zone are plotted using white squares (Fig. 4a). Although the plotting scheme we employed (Fig. 4a) does not distinguish between foliations that predate and accompanied emplacement of the Western Fiordland Orthogneiss, the scheme illustrates that the western domain preserves the oldest structures in the section. The ages of structures that define the foldand-thrust zone at Caswell Sound are constrained by several structural and metamorphic relationships in the contact aureole of the Western Fiordland Orthogneiss. The foliations that define the thrust zones all display subsolidus textures and cut the uppermost contact of the batholith. These characteristics indicate that the thrusts and the basal shear zone evolved after the batholith had cooled and crystallized. However, the thrust zones within 500 m of the Western Fiordland Orthogneiss contact also contain high-grade garnet-biotite-potassium feldspar granulite facies assemblages, whereas farther from the contact (>500 m) they contain chlorite-epidote amphibolite facies assemblages. Daczko et al. (2002) described these mineral assemblages in detail

MID-LOWER CRUSTAL ATTACHMENT ZONE

and showed that they reflect a temperature gradient of 700-800 °C within the contact aureole and 550-600 °C outside it. These observations suggest that the thrust zones initially formed within the thermally softened aureole of the batholith and continued to evolve as batholith cooled. The high pressures (7-9 kbar, Fig. Ib) and temperatures recorded by the mineral assemblages of the Caswell thrust zone and their contractional style indicate that they formed prior to the onset of regional extension at ~ 108 -105 Ma (Daczko et al 2002). Using this relationship and the fact that the thrust zone deforms the margins of the Western Fiordland Orthogneiss, the Caswell thrust zone must have evolved during the same time interval (116-105 Ma) as the George Sound shear zone and the Indecision Creek complex shear zones (ages from Klepeis et al. 2004). These features exhibit identical metamorphic grade and identical crosscutting relationships with respect to the Western Fiordland Orthogneiss. This interpretation also is consistent with the gradual merging of the George Sound shear zone with the basal shear zone of the Caswell thrust zone. The structural relationships we have outlined show a remarkable variability with depth in the crustal section (Fig. 3a, b). These relationships combined with good age control at the regional scale allowed us to investigate variations in the style of strain partitioning with depth for different time periods. Here, we define three stages in the magmatic and structural evolution of the Fiordland section: 1

the injection of mafic-intermediate magma of the Western Fiordland Orthogneiss batholith into a lower crust during the interval 126120 Ma, the partial melting of mafic host rocks below the batholith and its aureole (m, Fig. 2) accompanying this magmatism; 2 the mobilization and extraction of lower crustal melts during deformation that occurred as the batholith cooled to subsolidus temperatures (120-116 Ma); 3 the development of the steep upper amphibolite facies fabrics that define the Indecision Creek and George Sound shear zones and the Caswell thrust zone (116-105 Ma). The second and third stages are bracketed by the age of the Western Fiordland Orthogneiss and the youngest age of lower crustal fabrics that formed prior to the Mid-Cretaceous exhumation of the belt (~ 108-105 Ma). These age intervals allow for diachronous activity within the belt.

213

Stage 1: Mafic-intermediate magmatism and the partial melting of lower crust During the interval 126-120 Ma the Western Fiordland Orthogneiss was emplaced into lower crust composed of older (>126Ma) rocks of the Median Tectonic Zone and Palaeozoic margin of Gondwana. This magmatism resulted in a > 3000 km2 batholith that was at least 10km thick. The first phases of the batholith were gabbroic with minor ultramafic compositions; later phases were dominated by coarsegrained diorite. In some places, diffuse, undulate contacts between the gabbro and slightly younger diorite dykes suggest that these two phases mingled while still in a semi-molten state (Fig. 5a). The lower contact of the batholith at Mount Daniel (MD, Fig. 2) preserves a 200-500 m thick banded igneous complex. In this zone, sheets of tonalite, trondhjemite and hornblende gabbro display mutually crosscutting relationships and are complexly interfolded (Fig. 5b). Thin (<0.5 m) tonalite sheets display undulate contacts with slightly more mafic sheets reflecting injection into an incompletely crystallized host. Discordant amphibolite dykes with sharp, straight contacts cut some of these tonalite sheets but also are cut by veins and dyke apophases that originate from the surrounding tonalite host. These mutually crosscutting relationships indicate the simultaneous injection of tonalitic and more mafic phases. Many of the folds that occur within the banded igneous complex at the base of the Western Fiordland Orthogneiss preserves relationships that suggest that deformation occurred while the rocks were still in a partially molten state. Discordant patches of leucosome occur within the hinges of folded intrusions and are slightly folded (Fig. 5c). This texture suggests that the migration of melt, now represented by leucosome, occurred during folding of a partially molten tonalite host. Tightly folded and highly attenuated tonalite intrusions also are interfolded with slightly younger, less deformed tonalite sheets that cut them (Fig. 5d). The highly attenuated, stretched limbs of these folded layers are surrounded by leucosome that is unfoliated despite the evidence of high strains (Fig. 5d). Coarse biotite in the mafic parts of these layers forms radial and misaligned patterns. These relationships indicate that folding coincided with the periodic emplacement of the sheeted intrusions and suggest that the deformation occurred prior to full crystallization of the tonalite. In some areas near the base of the Western Fiordland Orthogneiss the surfaces of folded

214

K. A. KLEPEIS ETAL

Fig. 5. (a) The main gabbroic and dioritic phases of the Western Fiordland Orthogneiss batholith. Gabbro (dark colour) was invaded by slightly younger diorite (light colour). Note diffuse, undulate contacts suggesting magma mingling in a semi-molten state, (b) A 200-500 m thick, banded igneous complex located at the base of the Western Fiordland Orthogneiss at Mount Daniel (location shown in Fig. 2). Sectional view to the SW. Note the interfolding of tonalite, trondhjemite and gabbroic sheets. See text for discussion, (c) Discordant leucosome patches (L) in an incompletely crystallized tonalite host. Leucosome is located within the hinge of a complexly folded and stretched mafic amphibolite dyke (dark layers). The leucosome is unfoliated and slightly folded. Texture is interpreted to reflect the migration of melt now represented by leucosome during deformation of a partially molten tonalite host, (d) Sectional view to the SW. Tightly folded mafic (dark colour) layer in a tonalite host at the base of the Western Fiordland Orthogneiss at Mount Daniel. The folded layer is surrounded by leucosome (white). Despite the evidence for isoclinal folding, the tonalite host displays no microstructural evidence of subsolidus recrystallization near the stretched lower limb. Note that the folded layer and its tonalite host are truncated by a younger, less deformed banded tonalite intrusion at the top (arrows). Texture is interpreted to reflect deformation while the tonalite host was in a partially molten state. See text for discussion.

layers display SW-plunging hornblende and biotite mineral lineations that parallel the axes of the folds. These structures also parallel SWplunging hornblende mineral lineations observed elsewhere in the western domain (Fig. 4b). The sense of asymmetry displayed by the crosscutting sheets and the sheared out limbs of folds at Mount Daniel (MD, Fig. 2) suggests a topto-the-NE shear sense parallel to this SWplunging lineation. This asymmetry, the evidence of high strains and evidence of deformation while in a partially molten state suggest that this zone represents a melt-enhanced shear zone that accompanied emplacement of the Western Fiordland Orthogneiss (Fig. 3b). We recognized two generations of folds at the base of the Western Fiordland Orthogneiss at

Mount Daniel. The first generation includes the melt-enhanced folds that only occur within the 200-500 m thick basal shear zone. The second generation produced recumbent folds that refold the first generation. However, unlike the first generation, these second folds deform all units within the Western Fiordland Orthogneiss and underlying Milford Gneiss. These second-generation folds also display similar orientations and styles as those that formed within the western margin of the Indecision Creek shear zone near Mount Edgar (ME, Fig. 2). The deformation that produced these folds reoriented the sheeted intrusions but did not transpose the older fabrics. A penetrative axial planar foliation is lacking in most areas. This lack of penetrative deformation

MID-LOWER CRUSTAL ATTACHMENT ZONE

during folding preserved the migmatitic textures we describe in this section. Lastly, a few thin (< 10 m wide), mylonitic shear zones (Fig. 7d) parallel the axial planes of the second-generation folds. These shear zones display microstructural evidence that plagioclase grain sizes were reduced during subsolidus dynamic recrystallization. Some coarse plagioclase displays core-mantle structures and deformation bands composed of subgrains inside larger grains. On the basis of these relationships and similarities in structural style, we interpret these features as having formed during evolution of the Indecision Creek shear zone. Below and near the contacts of the Western Fiordland Orthogneiss, Palaeozoic gneiss and mafic rocks of the Arthur River complex preserve evidence of partial melting. Migmatite (m, Fig. 2) is best preserved in the western domain where delicate leucosome structures and granulite facies metamorphic mineral assemblages avoided most of the recrystallization and transposition that accompanied development of the Indecision Creek shear zone. The ~120 Ma metamorphic rims on zircons from migmatite and granulite facies mineral assemblages below the Western Fiordland Orthogneiss also support the interpretation that partial melting coincided with a thermal pulse that accompanied batholith intrusion. Field data show that the spatial distribution of rocks that partially melted was heterogeneous. Near the top of the batholith at George Sound, giant rafts of metapelitic host rock up to 4 km wide are enclosed by diorite (Figs 2 & 3a). These rafts preserve diatexite within 200500 m of the Western Fiordland Orthogneiss. Mineral assemblages in the diatexite, including garnet, biotite, hornblende and plagioclase that replace older staurolite- and kyanite-bearing assemblages, record the depth (25-30 km) of magma emplacement and partial melting at this locality (Table 1). At Caswell Sound migmatite was deformed and recrystallized by the deformation that produced the Caswell thrust zone (see also Daczko et al., 2002). In contrast to the narrow zones of partially melted rock near the top of the Western Fiordland Orthogneiss, migmatite formed in the Arthur River complex up to 10 km below the batholith north of Mount Daniel (m, Fig. 2). Daczko et al (200Ib) recognized that fluidabsent partial melting in the mafic parts of the Arthur River complex was patchy and involved the decomposition of hornblende at T > 750 °C. Piston-cylinder experiments conducted at the University of Vermont (Antignano et al. 2001; Antignano 2002) on a natural unmelted sample

215

of metadiorite from the Pembroke Granulite confirmed this relationship. These experiments produced two results that are of importance to the study of partial melting in the Arthur River complex during batholith emplacement. First, in accordance with field observations, they showed that at T = 850 °C biotite undergoes melting in the absence of free water followed by the reaction of hornblende and clinozoisite to form garnet plus melt as reaction products (Antignano et al 2001; Antignano 2002). Second, they showed that the melt fractions produced during melting remained low (<10%) at all temperatures up to T = 975 °C. This latter result may explain the low percentage of leucosome we observed in mafic Orthogneiss below the Western Fiordland Orthogneiss compared to the much higher melt fractions observed in migmatitic paragneiss near the top of the batholith (see also Klepeis et al 2003). Unlike the relationships we described for the basal shear zone at Mount Daniel, many areas that preserve migmatite below the Western Fiordland Orthogneiss do not display evidence of high-temperature ductile deformation during partial melting. This suggests that during the early stages of magmatism most of the deformation was partitioned along the basal contact of the batholith where a strength contrast occurred between partially molten rock and the older gneisses of the Arthur River complex. Stage 2: Melt segregation and transfer mechanisms Migmatitic rocks located structurally below the Western Fiordland Orthogneiss contain features that suggest that several different mechanisms helped to segregate and transport partial melts through the lower crust. In this section we describe two examples that illustrate how combinations of melt-induced fracture networks and ductile deformation in the Indecision Creek shear zone aided melt transfer during and after batholith emplacement. Melt-induced fracture propagation One of the best-exposed examples of migmatite in northern Fiordland is located in the Pembroke Granulite north of Milford Sound (P, Fig. 2). Here, dioritic gneiss displays a penetrative foliation composed of aggregates of hornblende, clinopyroxene, orthopyroxene, clinozoisite, plagioclase and small amounts of biotite and quartz. Leucosome in these rocks occurs as trains or patches of euhedral garnet partially

216

K. A. KLEPEIS ETAL

surrounded by aggregates of equidimensional plagioclase. The leucosome forms discontinuous lenses that are drawn out parallel to the foliation. However, there is no evidence of ductile deformation accompanying partial melting at this locality. These textures provide evidence that some of the original melt exploited and migrated along foliation planes following partial melting of the lower crust. In all areas of the Pembroke Granulite, the percentage of leucosome is low (<10%) and there is no disruption of stromatic layering. These observations support our interpretation that the volume of melt produced during partial melting of hornblende-bearing gneiss remained relatively low (<10%). They also suggest that the segregation of melt from migmatitic source rocks was efficient. We observed stromatic leucosome feeding laterally into larger discordant veins that cut the foliation at high angles (arrow in Fig. 6a). Garnet trains in these discordant veins are more extensively developed than in other areas of leucosome and are linked together into a continuous septum that is completely enclosed by coarse plagioclase (Fig. 6a). These features suggest that original partial melt was efficiently extracted from the migmatite by arrays of discordant veins. The veins that cut discordantly across foliation in the Pembroke Granulite are some of the most striking and penetrative features of these exposures. A regular lattice pattern consisting of steep orthogonal veins cuts across all rock contacts regardless of lithological composition (Fig. 6b, c). Some of these veins display en echelon patterns and curved, tapered tips (Fig. 6d). None of the veins display sigmoid shapes, sheared boundaries, or other features that typically indicate opening during deformation in shear zones. The curvature of some vein tips in the bridge between parallel but offset veins (Fig. 6d) is characteristic of a stress field modified by the internal fluid pressure of opening veins (Ramsay & Lisle 2000, pp. 766-767). These features are typical of en echelon tension gashes that form in regions of brittle deformation and high strain rates and provide strong field evidence that fracture propagation and dyking aided melt transfer through parts of the lower crust. Across contacts between the migmatitic diorite gneiss and adjacent, non-migmatitic gabbroic gneiss, a physical connectivity exists between leucosome in the migmatite and veins in the metagabbro (Fig. 6b). This physical connectivity revealed how migrating partial melts interacted chemically with gabbroic gneiss during their migration. Over distances of a few centimetres, leucosome in stromatic migmatite

changes into thin (2-5 cm) planar veins that are surrounded by dehydration zones in adjacent gabbroic gneiss (Fig. 6b-d; see also Daczko etaL 2001b; Klepeis et al 2003, fig. 3). These dehydration zones record the recrystallization of hornblende-bearing assemblages to garnet granulite at conditions of T> 750 °C and P = 1316 kbar (Table 1; Clarke et al 2000). Early theories suggested that the streaming of a CO2-rich fluid through fracture and vein networks caused the dehydration (Blattner 1976; Bradshaw 1989). However, the garnet granulite dehydration zones only occur in the gabbroic gneiss and are physically continuous with leucosome in the migmatitic diorite gneiss. These observations suggest that the dehydration of the metagabbro reflected migrating water-poor melt sourced from the migmatitic dioritic gneiss. On the basis of vein morphology, physical connectivity with migmatite and trace element data, Daczko et al (2001^) concluded that the veins and leucosome were produced by fracturing induced by a positive volume change during fluid-absent melting of the dioritic gneiss. The partial melting was inferred to have been controlled by the decomposition of hornblende and clinozoisite (Daczko et al 200\b). This mechanism of fracturing is similar to that proposed by Clemens & Mawer (1992), Connolly et al (1997) and Roering et al (1995), where high volumetric strain rates and a solid framework in host rock leads to the development of high pore fluid pressures in melt pockets due to local reaction. The elevated pressures lower effective normal stresses and eventually induce microfractures that propagate until fluid pressure drops below rock strength. The progressive migration of the melt maintains high fluid pressures and promotes continued microcrack propagation (Barker 1990). The hypotheses of melt-induced fracturing and the dehydration of host gneiss as a result of melt migration were tested for the Pembroke Granulite locality using partial melting experiments. Using a natural, unmelted sample of the dioritic gneiss, these experiments (Antignano et al 2001; Antignano 2002) established that the dilatational strains associated with melting involving the reaction of hornblende and clinozoisite were high enough to fracture matrix feldspar and quartz (see also Klepeis et al 2003, fig. 3B). In addition, the experiments showed that calculated water activities for these melts are low enough (0.39 to 0.12) to cause dehydration (Antignano et al 2001; Antignano 2002). The combined field, petrological and experimental data support the interpretation that fracture networks aided melt segregation in

MID-LOWER CRUSTAL ATTACHMENT ZONE

217

Fig. 6. (a) Stromatic migmatite in metadioritic host gneiss of the Pembroke Granulite (location shown in Fig. 2). Note foliation-parallel leucosome (arrow) that feeds a steep, discordant vein containing a continuous train of euhedral garnet to the right of the scale, (b) Structural form map showing the physical links that occur between garnetbearing leucosomes in migmatitic dioritic gneiss, and garnet- and clinopyroxene-bearing dehydration zones that surround tensile fractures in gabbroic gneiss (see also Daczko et al 200\b). (c) Tensile fracture arrays surrounded by dehydration zones form steep orthogonal arrays that cut all lithological contacts in the Pembroke Granulite. Host is gabbroic gneiss. Ellipse shows hammer for scale. View shows a bench-like surface where hammer is perched on a vertical wall above a horizontal surface. These veins and dehydration zones are defined by a central garnet-bearing leucosome surrounded by a halo (light grey lines) composed of garnet-clinopyroxene symplectite. (d) Tracing of a photograph showing the geometry of en echelon veins in metagabbroic host in the Pembroke Granulite. The veins show stepped tapered tips with a slight degree of curvature that is characteristic of tension fractures. Irregular vein boundaries probably reflect recrystallization during granulite facies metamorphism that accompanied vein formation. Black dots inside veins represent garnet formed as a product of melting involving the decomposition of hornblende and clinozoisite. Shaded areas represent garnet- and clinopyroxene-bearing dehydration fronts surrounding the veins. See text for discussion.

the Pembroke Granulite and that melt migration was linked to dehydration in the surrounding gabbroic rocks. An alternative mechanism to the fracture propagation hypothesis is focused flow in zones of low fluid pressure without creating microfractures. Such a mechanism has been observed in migmatite located elsewhere in Fiordland (described below). The migration of melt along grain boundaries could explain the diffuse boundaries of leucosome in the stromatic migmatite and some veins, and the collection of leucosome around garnet along foliation planes. However, focused flow alone does not explain the observed vein geometries. The orthogonal vein arrays and the tapered tips of en echelon tension gashes are characteristic of fractures forming under conditions of high fluid pressure and low differential stress. In addition, the irregular boundaries of some veins could reflect

high-temperature recrystallization or alteration during the garnet granulite facies metamorphism that accompanied vein formation. The lack of any evidence of ductile deformation during the formation of the orthogonal vein sets and dehydration zones suggests that the migration of melt into areas of low fluid pressure (e.g. boudin necks) during such deformation can be ruled out for this site. The distinctive vein networks with planar garnet granulite dehydration zones occur within a huge area of the western domain. We traced these features from the Pembroke Granulite into the lower part of the Western Fiordland Orthogneiss (see also Oliver 1980). This pervasive development suggests that melt-enhanced fracture propagation and the development of steep, regionally extensive fracture networks were important mechanisms of melt transfer during and immediately after emplacement of

218

K. A. KLEPEIS ETAL.

the Western Fiordland Orthogneiss. However, melt migration in propagating fractures was not the only mechanism of melt segregation and transport. Below, we contrast the characteristics of migmatite in the Pembroke Valley locality with those that record ductile deformation during partial melting within the Indecision Creek shear zone and the contact aureole of the Western Fiordland Orthogneiss. Melt accumulation in ductile shear zones Migmatite that formed in the metapelitic rafts exposed at George Sound contain an abundance of leucosome that suggests that high melt fractions (>30%) were produced and controlled by the reaction of biotite and muscovite during melting. An inhomogeneous diatexite in a metapelitic host near the contact aureole of the Western Fiordland Orthogneiss contains interconnected pockets of leucosome between irregular, joined palaeosome (Fig. 7a). The size and abundance of leucosome increases toward the contact between the batholith and its host where leucosome feeds discordant dykes that were injected back into the dioritic parts of the Western Fiordland Orthogneiss. Within a few tens of metres of the batholith contact, disconnected rafts of host rock are surrounded by leucosome that infill boudin necks. Differential movement between the rafts relative to gneissic layering in host rock is evident by the discordant pattern of foliations preserved inside them. These relationships suggest that melt, now represented by leucosome, migrated along foliation planes to low-pressure sites. However, apart from the localized boudinage and rotated blocks, this site does not record pervasive ductile deformation such as that which produced the Indecision Creek and George Sound shear zones. At Camp Oven Creek (CO, Fig. 2), stromatic migmatite in a dioritic host contains folded leucosome that parallels a folded foliation (Fig. 7c). This foliation contains the assemblage biotite, hornblende, clinozoisite, plagioclase and quartz. A second set of leucosomes that is not folded forms spaced arrays that parallel the axial surfaces of the folds (Figs 7c & 9a). The folds plunge gently to the south and their axial surfaces parallel the steep NNE-striking foliation of the Indecision Creek shear zone that occurs farther east. However, at this locality the folds lack an axial planar foliation. In some parts of the Camp Oven Creek exposures, the steep axial planar leucosomes truncate and offset the folded leucosomes (Figs 7c & 9a). This relationship suggests that

melt represented by the folded leucosome did not feed the second set of leucosomes but were below the solidus during deformation. However, the occurrence of leucosome parallel to the axial planes of folds suggests that these younger structures reflect melt migration during deformation in the shear zone. East of Camp Oven Creek, the abundance of leucosome decreases, most likely reflecting an increase in strain from the margin toward the central parts of the Indecision Creek shear zone. This strain gradient and the steep fabrics of the Indecision Creek shear zone are described in the next section.

Stage 3: Evolving styles of deformation following magmatism and crustal melting In this section we use outcrop-scale data to define the sequential evolution of ductile structures within and below the Western Fiordland Orthogneiss following partial melting of the lower crust. The relationships we describe illustrate the evolution of a network of steeply and gently dipping shear zones that culminated in formation of the steep Indecision Creek and George Sound shear zones (Fig. 4a). Steeply dipping sinistral and dextral shear zones In the Pembroke Valley north of Milford Sound (P, Fig. 2), a series of nearly vertical l-3m wide dextral and sinistral shear zones deform all vein arrays and garnet-bearing dehydration zones in metagabbro and metadiorite. These shear zones, and the migmatite and garnet dehydration zones at this locality, escaped the intense deformation and recrystallization accompanying development of the Indecision Creek complex (Fig. 4a). Bradshaw (1990) reported similar narrow and steep shear zones along the eastern side of the Western Fiordland Orthogneiss east of Mount Daniel. The shear zones display two orientations. The sinistral set contains a mylonitic foliation that strikes to the east and dips steeply to the south. A penetrative, gently SW-plunging mineral lineation defined by attenuated clusters of amphibole and clinozoisite occurs on these foliation planes. Sense-of-shear indicators within this set include oblique foliations, asymmetric recrystallized tails on feldspar porphyroclasts and microfaulted garnet. These structures all record sinistral displacements parallel to the mineral lineation. The second set of minor shear zones is subordinate in size and abundance to the first set. This second set contains mylonitic foliation

MID-LOWER CRUSTAL ATTACHMENT ZONE

219

Fig. 7. (a) A diatexite formed within 200 m of the contact between pelitic schist and the Western Fiordland Orthogneiss at the northwestern end of George Sound (Fig. 2 shows location). Diatexite occurs within a pelitic raft enclosed by the main dioritic phase of the batholith. (b) The steep subsolidus fabric that forms part of the George Sound shear zone inside the Western Fiordland Orthogneiss at George Sound (ellipse shows pencil for scale). Sectional view of a vertical face. SE is to the left, (c) Stromatic migmatite in dioritic gneiss at Camp Oven Creek (CO, Fig. 2). Note folded foliation-parallel leucosome cut by discordant leucosome that parallels the axial planes of the folds. Texture is interpreted to represent the migration of melt, now represented by the axial planar leucosome, during folding. See text for discussion. View is of a horizontal surface. SE is to the left, (d) Gently dipping mylonitic shear zone that formed within the Western Fiordland Orthogneiss near Mount Daniel (MD, Fig. 2).

that strikes to the NW and displays shallow to moderate dips to the SW. A gently west- and NW-plunging amphibole and clinozoisite mineral lineation occurs on foliation planes. Sense-of-shear indicators within this subordinate set all show dextral displacements. The kinematic evolution of these sinistraldextral shear zones is described in detail by Daczko et al (200la). Many of the shear zones appear to have nucleated on the margins of steep veins and the garnet-bearing dehydration zones where grain size and compositional differences produced strength contrasts. The results of using deformed vein sets as strain markers suggests that they record subhorizontal arcnormal (NW-SE) shortening and arc-parallel (NE-SW) stretching at the deepest levels of the section. Analyses of the mineral assemblages that define foliation in both sets of shear zones,

including garnet, pyroxene, hornblende, plagioclase, rutile and quartz, suggest that they equilibrated at peak conditions of P = 14.0 ± 1.3 kbar and T= 676 ± 34 °C (Daczko et al. 2001 a). Gently dipping, layer-parallel shear zones The steep shear zones of the Pembroke Valley are all deformed and transposed by a series of gently SE-dipping shear zones that approximately parallel the dominant gneissic foliations of the western domain. Each shear zone displays a mylonitic foliation consisting of the assemblage hornblende, clinozoisite, garnet and biotite. Stretched aggregates of garnet and plagioclase and aligned hornblende form down-dip, SEplunging mineral lineations on foliation planes. Daczko et al (200la) inferred that the metamorphic mineral assemblages within these

220

K. A. KLEPEIS ETAL.

shear zones equilibrated at peak conditions of p= 14.1 + l.2kbarandr=674 ± 36°C. The zone of mylonite that occurs in the central part of each of the gently dipping shear zones is approximately 7-10m thick (Fig. 8a). These central mylonite zones, all of which are approximately parallel, are vertically stacked on top of one another at a spacing of approximately 50-100m. The total thickness of this shear zone stack is unknown, but available exposure indicates that it is at least 1 km thick. Between the stacked shear zones, thin (< 1 m thick) mylonitic shear bands dip to the SE and NW and sweep into parallelism with the central mylonite zones located above and below them (Figs 3b & 8b). These shear bands envelop asymmetric pods of coarse-grained metagabbro and metadiorite that form imbricated, antiformal stacks that also dip to the SE and NW (Fig. 8). The displaced pods display an older, recrystallized gneissic foliation that is truncated by the margins of the shear bands. Lineations on foliation planes within the shear bands display similar trends but steeper plunges than the lineations that occur in the central mylonite zones. The sense of shear in all of these mylonite zones and shear bands is top-to-the-NW regardless of the dip of mylonitic foliations (Fig. 8). The style of displaced, imbricated pods that override one another in these shear zones is diagnostic of layer-parallel shortening and thickening in zones of contraction. Forward models of these geometries suggest that the antiformal style results when the successive stacking of displaced wedges during progressive deformation uplifts earlier-formed wedges and causes them to rotate (Ramsey & Huber 1987, p. 527). On the basis of these geometries and kinematics, we interpret these shear zones as a lower crustal duplex that records arc-normal or oblique (NWSE) shortening and vertical (layer-perpendicular) thickening of the crust below the Western Fiordland Orthogneiss. Daczko etal (2001a) suggested that, because the thrust zones and the sinistral shear zones described earlier both involved subhorizontal shortening at high angles to the MTZ arc, these structures reflect a partitioning of arcparallel and arc-normal components of oblique convergence onto sinistral strike-slip shear zones and ductile thrusts, respectively. Steep Indecision Creek and George Sound shear zones The steep foliation of the Indecision Creek shear zone is defined by flattened aggregates of garnet, hornblende, paragonite, biotite, clino-

zoisite, plagioclase and quartz. Hornblende and biotite mineral lineations on foliation planes display variable plunges to the south and SW (Fig. 4c). Stretched garnet and boudinage of amphibolite layers indicate that these mineral lineations represent a true stretching direction. Daczko (2001) showed that metamorphic conditions accompanying this deformation involved cooling from 750-800°C to 650-700°C with no corresponding change in pressure (stage 3 cooling, Table 1). This result is consistent with our results indicating that the Indecision Creek shear zone cuts across and transposes all suprasolidus fabrics on the eastern side of the Western Fiordland Orthogneiss (Fig. 4a). Along the western boundary of the shear zone, foliations of the western domain are deformed into a series of overturned, west-verging folds (Figs 3b & 4a) that plunge gently to moderately to the south and SW. We described earlier in this chapter how the geometry of these folds helps to define a regional strain gradient that increases from NW to SE into the Indecision Creek shear zone. Fold tightness (interlimb angle) and the dip of axial surfaces increase from NW to SE. Zones of high strain display tight to isoclinal, steeply plunging folds. In these areas, fold hinges and isolated pods that contain older foliations are enveloped and transposed parallel to a steep NNE-striking foliation (Figs 9b & lOb). The pods that are enveloped by the steep foliation occur at all scales. The largest of them preserves the thrust zones, migmatite and garnet dehydration zones of the Pembroke Granulite. This steep foliation is locally mylonitic (Fig. lOa) and parallels the boundaries of the Indecision Creek shear zone (Fig. 4a). Outcrop-scale sense-of-shear indicators occur mostly in zones of intermediate or low strain where the degree of transposition is less than in the high-strain areas. These low- to intermediate-strain zones occur mainly along the western boundary of the shear zone. Exceptions to this relationship occur in zones of mylonite, where sense-of-shear indicators are abundant (e.g. Fig. lOb). Sense-of-shear indicators include asymmetric hornblende and clinozoisite fish, asymmetric tails of biotite and hornblende on garnet porphyroblasts, asymmetric boudinage and minor shear zones. These indicators provide evidence of oblique sinistral displacements parallel to the moderately south- and SW-plunging mineral lineations that occur on the western side of the shear zone (Figs 4a & lOb). This sense of shear is consistent with the arc-parallel sinistral displacements recorded by steep minor shear zones of the Pembroke Valley.

MID-LOWER CRUSTAL ATTACHMENT ZONE

221

Fig. 8. (a) Sectional view of an outcrop-scale thrust duplex that deforms the Pembroke Granulite. Arrows show relative senses of displacement on mylonitic shear zones. SE is to the left, (b) Cross-sectional sketch of the shear zone forming the part of the duplex in (a). Black lines are foliation traces. Note imbricated asymmetric pods surrounded by mylonitic foliations that define a vertically thickening shear zone.

In addition to a dominantly sinistral shear sense, the involvement of folds at all scales and their change in geometry from NW to SE into the Indecision Creek shear zone provide evidence of a component of shortening across the shear zone. The steepening of fold axial surfaces that parallel the shear zone boundaries with increasing strain is consistent with arc-normal contraction. This interpretation also is in good agreement with data from both the sinistral and dextral shear zones and the ductile thrust zones exposed in the Pembroke Valley. Finally, changes in the plunge of hornblende mineral lineations on steep foliation surfaces with increasing strain (i.e. from NW to SE across the 1-5 km wide transitional zone) provide additional kinematic information. The plunges of these mineral lineations gradually steepen from gently and moderately SW-

plunging to near-vertical in high-strain zones (Fig. 4c). The migration of these lineations toward the dip-line of a vertical shear zone with increasing strain indicates that the shear zone was stretching or thickening in this downdip direction. The results of forward modelling of lineation and foliation trends in threedimensional shear zones by Lin et al (1998) and Jiang & Williams (1998) illustrate this relationship. In the reference frames provided by both the older gneissic foliations of the western domain and the steep boundaries of the shear zone, this result indicates that the Indecision Creek shear zone involved a component of vertical thickening. This interpretation is compatible with both a sinistral shear sense and subhorizontal shortening at high angles to the shear zone boundaries. Therefore, we conclude that this shear zone records subhorizontal NW

222

K. A. KLEPEIS ETAL.

Fig. 9. Structural form maps illustrating variations in the degree of transposition in the Indecision Creek shear zone, (a) The same migmatitic metadiorite as shown in Fig. 7c. Lightly shaded areas represent leucosome in stromatic migmatite. Darkly shaded areas represent deformed mafic dykes of the Milford Gneiss that intrude Palaeozoic host gneiss. Sketch location is at Camp Oven Creek (CO, Fig. 2) and lies within the transitional zone between the western domain and the Indecision Creek shear zone. Note folded, foliation-parallel leucosome cut by leucosome that parallels the axial planes of the folds. Axial planar leucosome parallels mylonitic foliation in the Indecision Creek shear zone, (b) Structural form map of a zone of high strain within the Indecision Creek shear zone located in Selwyn Creek Gneiss (Fig. 2). Foliation exhibits same orientation as axial planar leucosome shown in (a). See Figure 4c for orientation data. Note that the dominant foliation of the shear zone envelops gabbroic pods (darkly shaded) that preserve folds of an older foliation. Outside the pods the older foliation is completely transposed. Pods are surrounded by leucosome interpreted to reflect partial melt (lightly shaded). Random hatch pattern represents a diorite dyke. See text for discussion.

to SE arc-normal shortening and sinistral arcparallel displacements. The kinematic evolution of the George Sound shear zone is less well known than that of the Indecision Creek shear zone. However, numerous similarities exist between these two shear zones. Tight, upright, south-plunging folds of dykes (Fig. 5b) display geometries that are similar in style and orientation to the folds of

the Indecision Creek shear zone (Fig. 7b). Like in the Pembroke thrust zone near Milford Sound (Fig. 3b), gently dipping mylonitic shear zones deform older gently dipping fabrics inside the Western Fiordland Orthogneiss. The steep upper amphibolite facies fabrics of the George Sound shear zone also cut and transpose all older fabrics inside the Western Fiordland Orthogneiss. These steep foliations

MID-LOWER CRUSTAL ATTACHMENT ZONE

223

Fig. 10. (a) The deflection and transposition of a gneissic foliation (lower part) parallel to a steep mylonitic foliation (upper part) in the Indecision Creek shear zone, (b) Sketch of an isoclinal fold of a pegmatite that cuts an older gneissic foliation. Box shows location of photograph in (a). This style of gneissic foliation in lenses that are enveloped by mylonitic foliation occurs at all scales within the Indecision Creek shear zone. Sketch and photograph are from a dislodged exposure, so that foliations are not in their true orientation.

envelop lenses of diorite that preserve undeformed dykes and unrecrystallized magmatic textures. On the basis of similar structural styles, age and metamorphic grade, we suggested that the Indecision Creek shear zone formed during the same contractional event as the George Sound shear zone (see also section entitled Space-time correlation of high-grade fabrics).

Discussion: Structural evolution of a mid-lower crustal attachment zone The relationships we describe in this chapter show that contractional deformation following Early Cretaceous batholith emplacement and crustal melting produced a heterogeneous

network of steep and flat shear zones between 25 and 50 km depth. In this section we show how this crustal structure and the kinematic relationships that developed during the contraction define a mid-lower crustal attachment zone that linked arc-normal displacements in a mid-crustal fold-and-thrust belt to complexly deforming subvertical shear zones in the lowermost crust. During the period 126-120 Ma, meltenhanced shear zones formed in narrow zones at the upper and lower contacts of the Western Fiordland Orthogneiss (Stage 1, Fig. 11 a). Below the batholith, metamorphic data indicate that the thermal pulse accompanying this magmatism affected up to 10 km of crust below the batholith. However, little penetrative ductile

224

K. A. KLEPEIS ETAL.

deformation appears to have accompanied the earliest stages of migmatite formation and melt-induced fracturing in most areas of the Arthur River complex. These observations suggest that magma and steep temperature gradients associated with emplacement of the Western Fiordland Orthogneiss created large strength contrasts that focused deformation inside and at the margins of the batholith during this period. As the batholith cooled and partial melts were extracted efficiently from beneath it (120116 Ma, Stage 2, Fig. lib), the style of deformation at the deepest levels of the section changed. The melt-enhanced shear zone that localized at the base of the Western Fiordland Orthogneiss was abandoned and a series of deformations evolved below it during the period 116-105 Ma (Stage 3, Fig. lie). The largest of the shear zones that formed during this period cut across the lower and central parts of the batholith and at least the ~10km of crust beneath it (Figs 3b & 4a). The Indecision Creek and George Sound shear zones represent the last phases in this series of contractional deformations affecting the lower crust following batholith emplacement. The earliest stages of this contraction produced a lower crustal thrust system characterized by simultaneous or nearly simultaneous motion on a series of interconnected steep and flat shear zones. Similar networks of flat and steep shear zones have been described by Karlstrom & Williams (2002). In addition, the Fiordland exposures illustrate a cyclic alternation between the development of steep sinistral shear zones at different times and the formation of flat ductile thrust zones at different depths in the crust (Fig. lie). However, despite the crosscutting relationships, there are numerous kinematic similarities between these different phases of deformation. All of the deformations are consistent with subhorizontal shortening at high angles to the NNE trend of the arc (MTZ). The last two phases of deformation, represented by the Pembroke thrust zone and the Indecision Creek shear zone, both record vertical (layerperpendicular) thickening as well as subhorizontal shortening. Finally, the steep minor shear zones of the Pembroke Valley and the Indecision Creek shear zone both record sinistral arc-parallel displacements at slightly different times. An entirely different structural style and evolution than that which occurred below the batholith characterizes the region of the crust located at the upper margin and above the batholith (Figs 3 & lie). At Caswell Sound, the ductile thrust system that localized at the upper contact of the

Western Fiordland Orthogneiss was not abandoned but continued to evolve as the batholith cooled below the solidus to temperatures of T< 800 °C after 116 Ma (Klepeis et al 2004). Our analysis of exposure below the Caswell thrust zone also indicates that the steep foliations of the George Sound shear zone flatten and merge with the 1 km thick shear zone that forms the base of the Caswell thrust zone (Figs 3 & 4a). Yet, despite their different styles and orientations, both the Caswell thrust zone and the steep shear zones located structurally below it also record subhorizontal shortening at high angles to the arc and vertical thickening. These relationships suggest that, despite the variability in size, style and orientation of the shear zones located above and below the batholith, all phases of deformation during the period 116-105 Ma are consistent with arc-normal shortening, vertical thickening and a preferential partitioning of arc-parallel displacements on steep NNE-striking surfaces. These kinematic links between different crustal levels define a transfer zone between westdirected shortening in the Caswell fold-andthrust zone and oblique sinistral displacements with a component of arc-normal shortening in the Indecision Creek shear zone. Fabric correlations based on regional-scale crosscutting relationships and U-Pb data outlined earlier in this chapter suggest that these systems evolved during the same 116-105 Ma interval. Although different mechanisms apply, simultaneous movement above and below the Caswell thrust zone is required to accommodate shortening at different crustal depths. We define the basal shear zone that underlies the Caswell thrust zone as an attachment zone that linked simultaneous deformation on an interconnected network of flat and steep shear zones. The 1 km thick shear zone that underlies the Caswell belt divided the crust into domains exhibiting widely different styles and physically attached the thrusts to the steep, complexly deforming shear zones below it. The middle crust above the attachment zone is more partitioned and segmented than the penetrative 1015 km wide zones of ductile deformation that define the shear zones below it. Nevertheless, despite the different styles of deformation, structures above and below the attachment zone are compatible with arc-normal shortening that was accommodated on surfaces of different orientation. In addition, the steep shear zones that formed at the deepest levels of the section accommodated a component of arc-parallel displacements on steep NNE-striking surfaces. This latter relationship leads to the prediction that steep, arc-parallel strike-slip faults probably

Fig. 11. Three stages in the evolution of the Fiordland crustal section, (a) Stage 1 (126-120 Ma): The injection of mafic-intermediate magma of the Western Fiordland Orthogneiss batholith into a layered lower crust, (b) Stage 2 (120-116 Ma): The extraction of partial melts from below the batholith via regional networks of veins and dykes, (c) Stage 3 (116-105 Ma): Contractional deformation followed batholith emplacement and partial melting of the lower crust. Contraction culminated in development of the Indecision Creek and George Sound shear zones that cut across the batholith and merge with a mid-crustal fold-and-thrust zone located at the top of the batholith. The thrusts root into a 1 km thick shear zone that is defined in the text as an attachment zone. This attachment zone links the ductile thrust zone above (bold black lines) to the steep foliations of the Indecision Creek shear zone below. Data presented in the text indicate that these features evolved during the same time period after the batholith cooled and crystallized. See text for discussion.

226

K. A. KLEPEIS ETAL

formed west of the MTZ at upper crustal levels at this time. Conclusions The Fiordland belt records an Early Cretaceous history of mid-lower crustal magmatism, highgrade metamorphism and intense contractional deformation that occurred at palaeodepths of 25-50 km. From 126-120 Ma, a >10km thick batholith composed of mafic-intermediate magma was emplaced into the lower crust section. During magma emplacement, layer-parallel, melt-enhanced shear zones formed at the upper and lower boundaries of the batholith where strength contrasts were produced by magma and steep temperature gradients. Magmatism was accompanied by granulite facies metamorphism and the formation of migmatite up to 10km below the intruding batholith. Fluid-absent melting of mafic-intermediate gneiss was controlled mostly by the decomposition of hornblende + clinozoisite. Positive volume changes and high melt fluid pressures during melt production resulted in fracture networks that aided the segregation and transfer of melt through and out of the lower crust during the period 120116 Ma. Ductile deformation in steep shear zones also was important for moving partial melts vertically and horizontally through the lower crust. During the period 116-105 Ma, the batholith and rocks located below it cooled to temperatures of T < 800 °C, and differential shortening of the crust created a heterogeneous network of subvertical and subhorizontal shear zones. During the earliest stages of this deformation, these shear zones exploited the layered architecture of the middle and lower crust and separated it into structural domains exhibiting widely different structural styles. A ductile fold-and-thrust system formed in the middle crust (25-30 km palaeodepths), with garnet granulite and upper amphibolite facies thrust splays rooting down into a 1 km thick, nearly horizontal shear zone that localized at the uppermost contact of the batholith. This flat shear zone physically connected the mid-crustal fold-and-thrust zone to subvertical shear zones in the lower crust that are up to 15 km wide. The subvertical foliations in these steep shear zones cut across the lower contact of the batholith and merge smoothly with the subhorizontal shear zone in the middle crust. Fabric correlations based on regionalscale crosscutting relationships and U-Pb data strongly suggest that these systems evolved simultaneously or nearly so during the same 116-105 Ma interval. Simultaneous movement

above and below the mid-crustal fold-andthrust zone also is required to accommodate shortening at different crustal depths. This crustal structure and evidence of kinematic links above and below the flat shear zone at the top of the batholith define a midlower crustal attachment zone that accommodated differential displacements at different crustal depths. Together, this network of flat and steep shear zones accommodated mostly subhorizontal east-west, arc-normal shortening in the fold-and-thrust zone at mid-crustal depths and oblique sinistral displacements on steep, complexly deforming shear zones at lower crustal depths. The oblique-sinistral displacements also were accompanied by a component of vertical thickening and subhorizontal arc-parallel displacements on NNE-striking surfaces. These displacement patterns are consistent with oblique convergence along an Early Cretacous plate boundary located outboard of the early Mesozoic arc. We are grateful to Cheryl Waters, Olivier Vanderhaeghe and Basil Tikoff for suggestions on ways to improve the original manuscript. We also thank N. Daczko, I. Turnbull, N. Mortimer and A. Tulloch for helpful discussions during this study. J. Hollis, A. Claypool, S. Marcotte, W.C. Simonson and G. Mora-Klepeis also provided valuable assistance. We thank the Department of Land Conservation in Te Anau for permission to visit and sample localities mentioned in the text. This study was funded by a National Science Foundation grant to KAK (EAR0087323), an Australian Research Council grant to KAK and GLC (ARC-A10009053), and grants from the Geological Society of America and the University of Vermont.

References ADAMS, C.J. 1975. New Zealand potassium argon age list, 2. New Zealand Journal of Geology and Geophysics, 18, 443-467. ANTIGNANO, A., IV, 2002. Experimental constraints on granitoid compositions in convergent regimes: a geochemical study. MSc Thesis, University of Vermont. ANTIGNANO, A., IV, RUSHMER, T, DACZKO, N.R., CLARK, G.L., COLLINS, WJ. & KLEPEIS, K.A. 2001. Partial melting of a hornblende-biotiteclinozoisite bearing metadiorite: applications to the deep crust, Fiordland, New Zealand. Geological Society of America Abstracts with Programs, 33(6), 211. AXEN, G.J., SELVERSTONE, J, BYRNE, T & FLETCHER, J.M. 1998. If the strong crust leads, will the weak crust follow? GSA Today, 8, 1-8. BARKER, AJ. 1990. Introduction to Metamorphic Textures and Micro structures. Chapman and Hall, New York. BEAUMONT, C, JAMIESON, R.A., Nguyen, M.H. & LEE, B. 2001. Himalayan tectonics explained by

MID-LOWER CRUSTAL ATTACHMENT ZONE extrusion of a low-viscosity crustal channel coupled to focused surface denudation. Nature, 414, 738-742. BLATTNER, P. 1976. Replacement of hornblende by garnet in granulite facies assemblages near Milford Sound, New Zealand. Contributions to Mineralogy and Petrology, 55, 181-190. BLATTNER, P. 1978. Geology of the crystalline basement between Milford Sound and the Hollyford Valley, New Zealand. New Zealand Journal of Geology and Geophysics, 21, 33-47. BLATTNER, P. 1991. The North Fiordland transcurrent convergence. New Zealand Journal of Geology and Geophysics, 34, 543-553. BRADSHAW, J.D. 1989. Cretaceous geotectonic patterns in the New Zealand region. Tectonics, 8, 803-820. BRADSHAW, J.Y. 1985. Geology of the northern Franklin Mountains, northern Fiordland, New Zealand, with emphasis on the origin and evolution of Fiordland granulites. Unpublished PhD Thesis, University of Otago, Dunedin, New Zealand. BRADSHAW, J.Y. 1989. Origin and metamorphic history of an Early Cretaceous polybaric granulite terrain, Fiordland, southwest New Zealand, Contributions to Mineralogy and Petrology, 103, 346-360. BRADSHAW, J.Y. 1990. Geology of crystalline rocks of northern Fiordland: details of the granulite facies Western Fiordland Orthogneiss and associated rock units. New Zealand Journal of Geology and Geophysics, 33, 465-484. BRADSHAW, J.Y. & KIMBROUGH, D.L. 1989. Comment: Age constraints on metamorphism and the development of a metamorphic core complex in Fiordland, southern New Zealand. Geology, 17, 380-381. CLARKE, G.L., KLEPEIS, K.A. & DACZKO, N.R. 2000. Cretaceous high-P granulites at Milford Sound, New Zealand: metamorphic history and emplacement in a convergent margin setting. Journal of Metamorphic Geology, 18, 359-374. CLAYPOOL, A., KLEPEIS, K., DOCKRILL, B., CLARKE, G., ZWINGMANN, H. & Tulloch, A. 2002. Structure and kinematics of oblique continental convergence in northern Fiordland, New Zealand. Technophysics, 359, 329-358. CLEMENS, J.D. & MAWER, C.K. 1992. Granitic magma transport by fracture propagation. Tectonophysics, 204, 339-360. CONNOLLY, J.A.D., HOLNESS, M.B., RUBIE, D.C. & RUSHMER, T. 1997. Reaction-induced micro-cracking: an experimental investigation of a mechanism for enhancing anatectic melt extraction. Geology, 25,591-594. DACZKO, N.R. 2001. The structural and metamorphic evolution of Cretaceous high-P granulites, Fiordland, New Zealand. Unpublished PhD Thesis, University of Sydney, Australia. DACZKO, N.R., KLEPEIS, K.A. & CLARKE, G.L. 2001<2. Evidence of Early Cretaceous collisional-style orogenesis in northern Fiordland, New Zealand and its effects on the evolution of the lower crust, Journal of Structural Geology, 23, 693-713.

227

DACZKO, N.R., CLARKE, G.L. & KLEPEIS, K.A. 2001£. Transformation of two-pyroxene hornblende granulite to garnet granulite involving simultaneous melting and fracturing of the lower crust, Fiordland, New Zealand. Journal of Metamorphic Geology, 19, 549-562. DACZKO, N., KLEPEIS, K.A. and CLARKE, G.L. 2002. Thermomechanical evolution of the crust during convergence and deep crustal pluton emplacement in the Western Province of Fiordland, New Zealand, Tectonics, 21, 1-18. ECKERT, J.O., JR., NEWTON, R.C. & KLEPPA, O.J. 1991. The A// of reaction and recalibration of garnet-pyroxene-plagioclase-quartz geobarometers in the CM AS system by solution calorimetry. American Mineralogist, 76, 148-160. ELLIS, D.J. & GREEN, D.H. 1979. An experimental study of the effect of Ca upon garnet-clinopyroxene Fe-Mg exchange equilibria. Contributions to Mineralogy and Petrology, 71, 13-22. ELLIS, S., BEAUMONT, C., JAMIESON, R. & QUINLAN, G., 1998. Continental collision including a weak zone - the Vise model and its application to the Newfoundland Appalachians. Canadian Journal of Earth Science, 35, 1323-1346. FERRY, J.M. & SPEAR, F.S. 1978. Experimental calibration of the partitioning of Fe and Mg between biotite and garnet. Contributions to Mineralogy and Petrology, 66, 113-117. Fuis, G.S. & CLOWES, W.D. 1993. Comparison of deep structure along three transects of the western North American continental margin. Tectonics, 12, 1420-1435. GIBSON, G.M., & IRELAND, T.R. 1995. Granulite formation during continental extension in Fiordland. Nature, 375, 479-482. GIBSON, G.M., MCDOUGALL, I. & IRELAND, T.R. 1988. Age constraints on metamorphism and the development of a metamorphic core complex in Fiordland, southern New Zealand. Geology, 16, 405-408. GRAHAM, C.M. & POWELL, R. 1984. A garnethornblende geothermometer; calibration, testing, and application to the Pelona Schist, Southern California. Journal of Metamorphic Geology, 2, 13-31. HARRY, D.L., OLDOW, J.S. & SAWYER, D.S. 1995. The growth of orogenic belts and the role of crustal heterogeneities in decollement tectonics. Geological Society of America Bulletin, 107, 1411-1426. HILL, E.J. 1995. A deep crustal shear zone exposed in western Fiordland, New Zealand. Tectonics, 14, 1172-1181. HODGES, K.V. & SPEAR, F.S. 1982. Geothermometry, geobarometry and the Al2SiO5 triple point at Mt. Moosilauke, New Hampshire. American Mineralogist, 66, 1118-1134. HOLLIS, J.A., CLARKE, G.L., KLEPEIS, K.A., DACZKO, D.A. & IRELAND, T.R. 2002. Geochronology and geochemistry of high-pressure granulites of the Arthur River Complex, Fiordland, New Zealand: Cretaceous magmatism and metamorphism on the Palaeo-Pacific margin. Journal of Metamorphic Geology, in press.

228

K. A. KLEPEIS ETAL.

JIANG, D. & WILLIAMS, P.P. 1998. High-strain zones: McCuLLOCH, M.T., BRADSHAW, J.Y. & TAYLOR, S.R. a unified model. Journal of Structural Geology, 20, 1987. Sm-Nd and Rb-Sr isotopic and geochemical systematics in Phanerozoic granulites from 1105-1120. Fiordland, Southwest New Zealand. Contributions KARLSTROM, K.E. & WILLIAMS, M.L. 2002. Nature of to Mineralogy and Petrology, 97, 183-195. the middle crust - heterogeneity of structure and process due to pluton-enhanced tectonism: an MILLER, R.B. & PATERSON, S.R. 2001. Influence of lithological heterogeneity, mechanical anisotropy, example from Proterozoic rocks of the North and magmatism on the rheology of an arc, North American Southwest. In: BROWN, M & Cascades, Washington. Tectonophysics, 342, RUSHMER, T. (eds) Evolution and Differentiation 351-370. of the Continental Crust. Cambridge University MOLNAR, P., ANDERSON, H.J., et al. 1999. Continuous Press, Cambridge, in press. deformation versus faulting through the continental KlMBROUGH, D.L., TULLOCH, A.J., COOMBS, D.S., lithosphere of New Zealand. Science, 286, 516LANDIS, C.A., JOHNSTON, M.R. & MATTINSON, J.M. 519. 1994. Uranium-lead zircon ages from the Median Tectonic Zone, New Zealand. New Zealand Journal MORTIMER, N., TULLOCH, A.J., SPARK, R., WALKER, N., LADLEY, E., KIMBROUGH, D.L. & ALLIBONE, A.H. of Geology and Geophysics, 37, 393-419. 1999. Overview of the Median Batholith, New KLEPEIS, K.A., DACZKO, N.R. & CLARKE, G.L. 1999. Zealand: a new interpretation of the geology of Kinematic vorticity and tectonic significance of the Median Tectonic Zone and adjacent rocks. superposed mylonites in a major lower crustal Journal of African Earth Science, 29, 257-268. shear zone, northern Fiordland, New Zealand. MUIR, R.J., IRELAND, T.R., WEAVER, S.D. & Journal of Structural Geology, 21, 1385-1405. BRADSHAW, J.D. 1994. Ion microprobe U-Pb KLEPEIS, K.A., CLARKE, G.L., DACZKO, N, zircon geochronology of granitic magmatism RUSHMER, T., COLLINS, W. & TULLOCH, A. 2001. in the Western Province of the South Island, The effects of partial melting and magmatism in the New Zealand. Chemical Geology, 113, 171-189. lower crust on the evolution of continental lithosphere: results from Fiordland, New Zealand. MUIR, R.J., WEAVER, S.D., BRADSHAW, J.D., EBY, G.N. & EVANS, J.A. 1995. The Cretaceous Geological Society of America Abstracts with separation point Batholith, New Zealand; granitoid Programs, 33(6), 153. magma formed by melting of mafic lithosphere. KLEPEIS, K.A., CLARKE, G.L. & RUSHMER, T. 2003. Journal of the Geological Society, London, 152, Magma transport and coupling between defor689-701. mation and magmatism in the continental lithoMUIR, R.J., IRELAND, T.R., WEAVER, S.D., sphere. GSA Today, 13, 4-11. BRADSHAW, S.D., EVANS, J.A., EBY, G.N. & KLEPEIS, K.A., CLARKE, G.L., GEHRELS, G. & SHELLEY, D. 1998. Geochronology and geochemVervoort, J. 2004. Processes controlling vertical istry of a Mesozoic magmatic arc system, Fiordcoupling and decoupling between the upper land, New Zealand. Journal of the Geological and lower crust of orogens: results from Fiord, Society, London, 155, 1037-1053. New Zealand. Journal of Structural Geology, 26, NATHAN, S., THURLOW, C., WARNES, P. & 765-791. ZUCCHETTO, R. 2000. Geochronology Database KOHN, M.J. & SPEAR, F.S. 1990. Two new barometers for New Zealand Rocks, 2nd edn.: 1961-1999. Instifor garnet amphiboles, with applications to southtute for Geologic and Nuclear Science Report, 11,51. eastern Vermont. American Mineralogist, 75, NEWTON, R.C. & HASLETON, H.T. 1981. Thermo89-96. dynamics of the garnet-plagioclase-Al2SiO5KROGH, E.J. 1988. The garnet-clinopyroxene Fe-Mg quartz geobarometer. In: NEWTON, R.C., NAVgeothermometer: a reinterpretation of existing ROTSKY A. & WOODS B.J. (eds) Advances in experimental data. Contributions to Mineralogy Physical Geochemistry, 1. Springer Verlag, and Petrology, 99, 44-48. New York, pp. 131-147. LIN, S., JIANG, D. & WILLIAMS, P.P. 1998. Transpres- NEWTON, R.C. & PERKINS, D, 1982. Thermodynamic sion (or transtension) zones of triclinic symmetry: calibration of geobarometers based on the assemnatural example and theoretical modelling. In: blages garnet -plagioclase - orthopyroxene - (clinoHOLDSWORTH, R.E., STRACHAN, R.A. & DEWEY, pyroxene)-quartz. American Mineralogist, 67, J.F. (eds) Continental Transpressional and Trans203-222. tensional Tectonics. Geological Society, London, OLDOW, J.S., BALLY, A.W., AVE LALLEMANT, H.G. Special Publications, 135, 41-47. 1990. Transpression, erogenic float, and lithoMATTINSON, J.L., KIMBROUGH, D.L. & spheric balance. Geology, 18, 991-994. BRADSHAW, J.Y. 1986. Western Fiordland ortho- OLIVER, GJ.H. 1980. Geology of the granulite and gneiss: Early Cretaceous arc magmatism and amphibolite facies gneisses of Doubtful Sound, granulite facies metamorphism, New Zealand. Fiordland, New Zealand. New Zealand Journal of Contributions to Mineralogy and Petrology, 92, Geology and Geophysics, 1, 27-41. 383-392. POWELL, R. 1985. Regression diagnostics and robust MAYER, G., MAI, P.M., et al. 1997. The deep crust of regression in geothermometer/geobarometer calithe southern Rhine Graben: reflectivity and seismibration; the garnet-clinopyroxene geothermocity as images of dynamic processes. Tectonophymeter revisited. Journal of Metamorphic Geology, sics, 275, 15-40. 3,231-243.

MID-LOWER CRUSTAL ATTACHMENT ZONE POWELL, R. & HOLLAND, TJ.B. 1988. An internally consistent dataset with uncertainties and correlations: 3. Applications to geobarometry, worked examples and a computer program. Journal of Metamorphic Geology, 6, 173-204. RAMSAY, J.G. & HUBER, M.I. 1987. The Techniques of Modern Structural Geology, Volume 2: Folds and Fractures. Academic Press, New York. RAMSAY, J.G. & LISLE, RJ. 2000. The Techniques of Modern Structural Geology, Volume 3: Applications of Continuum Mechanics in Structural Geology. Academic Press, New York. ROERING, C., VAN REENEN, D.D., SMIT, C.A. &

Du TOIT, R 1995. Deep crustal embrittlement and fluid flow during granulite metamorphism in the Limpopo Belt, South Africa. Journal of Geology, 103, 673-686. ROYDEN, L. 1996. Coupling and decoupling of crust and mantle in convergent orogens: implications for strain partitioning in the crust. Journal of Geophysical Research, 101, 17679-17705. SUTHERLAND, R. & HOLLIS, C. 2001. Cretaceous demise of the Moa Plate and strike-slip motion at the Gondwana margin. Geology, 29, 279-282. TEYSSIER, C., TIKOFF, B. & WEBIR, J. 2002. Attachment between brittle and ductile crust at wrenching plate boundaries. EGS Stephen Mueller Special Publication Series, European Geophysical Society, 1, 119-144. TULLOCH, A.J. & KIMBROUGH, D.L. 1989. The Paparoa metamorphic core complex, New Zealand: Cretac-

229

eous extension associated with fragmentation of the Pacific margin of Gondwana. Tectonics, 8, 1217-1234. TULLOCH, A.J. & KIMBROUGH, D.L. 2003. Paired plutonic belts in convergent margins and the development of high Sr/Y magmatism: the Peninsular Ranges Batholith of California and the Median Batholith of New Zealand. In: JOHNSON, S.E. PATERSON, S.R., FLETCHER, J.M., GIRTY, G.H., KIMBROUGH, D.L. & MARTIN-BARAJAS, A. (eds), Tectonic Evolution of Northwestern Mexico and the Southwestern USA, Boulder, Colorado. Geological Society of America Special Paper, 374, 275-295. TULLOCH, A.J., IRELAND, T.R., WALKER, N.W. & KIMBROUGH, D.L. 2000. U-Pb Zircon Ages from the Milford Orthogneiss, Milford Sound, Northern Fiordland: Paleozoic Igneous Emplacement and Early Cretaceous Metamorphism. Institute for Geologic and Nuclear Science Report, 6, 17. TURNBULL, I.M. 2000. Geology of the Wakatipu Area. Institute of Geological and Nuclear Sciences 1:250 00 Geologic Map 18. Institute of Geological and Nuclear Sciences Ltd, Lower Hutt, New Zealand. WILLIAMS, J.G. & HARPER, C.T. 1978. Age and status of the Mackay Intrusives in the Eglinton-Upper Hollyford area. New Zealand Journal of Geology and Geophysics, 21, 733-742.

This page intentionally left blank

Attachment formation during partitioning of oblique convergence in the Ketilidian orogen, south Greenland K. J. W. MCCAFFREY1, J. GROCOTT2, A. A. GARDE3 & M. A. HAMILTON4* 1 Department of Earth Sciences, University of Durham, Durham, DHl 3LE, UK 2 Earth Sciences and Geography, Kingston University, Kingston upon Thames, Surrey, KT1 2EE, UK 3 GEUS, OsterVolgade 10, 1350 Copenhagen K, Denmark 4 Geological Survey of Canada, 601 Booth Street, Ottawa, ON, K1A OE8 Canada * Present address: Department of Geology, Earth Sciences, University of Toronto, 22 Russell St, Toronto, ON, M5S 3B1, Canada Abstract: Subhorizontal attachment zones provide coupling between lithospheric layers in orogenic belts. A mid-crustal attachment zone is exposed in the Palaeoproterozoic Ketilidian orogen, south Greenland, which formed as a result of north-directed oblique convergence at a cordilleran-type margin. Rifting (c. 2.1 Ga) and compressional deformation and magmatism (> 1850 Ma) on the continental margin was followed by an extended sinistral transpression from 1850 to 1730 Ma now separated into three episodes or peaks of activity. The first episode was focused on the back-arc region and was followed by the main arc construction phase during which transpression was partitioned into strike-slip and contraction components. Despite the longevity of this active margin system, individual tectonic events took place rapidly, e.g. development of fore-arc Dj-D 3 and accompanying high-temperature, low-pressure metamorphism took place over c. 12 Ma. We explain the fore-arc and batholith evolution by the upward migration of an underlying attachment structure through the upper crustal partitioned blocks. This migration may be attributed to an increase in the geothermal gradient accompanied by, or followed by, exhumation of the mid-crust. The partially molten, hence weak, attachment zone solidified and strengthened during cooling before emplacement of the post-orogenic rapakivi suite during the third distinct phase of mild sinistral transpression.

Oblique plate convergence forms orogenic belts that display complex three-dimensional transpressional architectures as shown by many field studies, predicted by analytical descriptions and exhibited by numerical and analogue experiments (Dewey 1975; Mackenzie & Jackson 1983; Richard & Cobbold 1990; Teyssier & Tikoff 1998). In the upper crust, partitioning of transpression produces coexisting orogennormal shortening and orogen-parallel strikeslip deformation (Fitch 1972; McCaffrey 1992; Molnar 1992; Tikoff & Teyssier 1994; Dewey et al 1998). The nature of this deformation partitioning is likely to change with depth due to the differing rheologies exhibited by the upper and lower crust, and the lithospheric mantle (Molnar 1992; Kohlstedt et al 1995; Roy den 1996; Mackenzie & Jackson 2002). The mid- to lower crust is typically composed of amphibolite to granulite facies rocks. In many exhumed orogenic belts these rocks display originally subhorizontal peak metamorphic fabrics

and evidence for high shear strains, e.g. sheath folds (Strachan et al. 1992; Northup & Burchfiel 1996). These shear zones have been termed 'basal detachments' or 'decollements' (Richard & Cobbold 1990; Oldow et al. 1990; Teyssier & Tikoff 1998), with implied decoupling between the upper crust and the mantle lithosphere. Recently, Tikoff etal. (2001) and Teyssier et al. (2002) have proposed the term 'attachment' for a subhorizontal shear zone across which there is coupling between lithospheric layers that display different rheological properties. Tikoff et al (2001) further suggest that these attachment zones are important for transmitting driving forces vertically within an orogenic belt and question the importance of horizontal or side-acting tectonic driving forces in orogenic belts. Attachment zones should be recognizable by 3D architectures and kinematic frameworks that are compatible with an obliquely convergent orogen, taking into account contrasting rheologies

From: GROCOTT, JL, MCCAFFREY, K. J. W., TAYLOR, G. & TIKOFF, B. (eds) 2004. Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 231-248. 0305-8719/04/$15 © The Geological Society of London 2004.

232

MCCAFFREY ET AL.

typical of different lithospheric levels. In this chapter, we examine an obliquely convergent orogenic belt, exposed at mid- to lower crustal levels, in order to assess whether its overall architecture is compatible with that of a partitioned transpressional orogen with an attachment structure. The Ketilidian orogen is dramatically exposed in the alpine topography of south Greenland, and a range of metamorphic grades suggest that different crustal levels are exposed. We provide a summary of the architecture and tectonic history of the Ketilidian orogen (Fig. 1), a type example of Palaeoproterozoic (c. 1800 Ma) continental growth by arc magmatism. It formed in response to convergence between an oceanic plate and an Archaean craton (Chadwick & Garde 1996; Garde et al 20020). Unlike many other Palaeoproterozoic orogens, primary structure is well preserved because the orogen escaped modification by collisional processes. The evolution of the Ketilidian orogen has been outlined elsewhere (e.g. Garde et al. 20020, b), but here the aim is to focus on the overall tectonic architecture of the orogen and to discuss the nature and timing of the main tectonothermal events that affected its component parts in the context of a preserved attachment structure.

(Berthelsen & Henriksen 1975, and references therein) (Fig. 1). In the northwestern Border Zone (NWBZ), supracrustal rocks in Midternaes and Grasnseland collectively form a minimum 5 km thick stratigraphic sequence that overlies peneplained Archaean basement and includes siliciclastic and calcareous rocks with chertpodded iron formation and overlying c. 4 km thick Sortis Group basic metavolcanic rocks. In the northeastern part of the orogen, the northeastern Border Zone (NEBZ) is exposed between Mogens Heinesen Fjord and Napasorsuaq Fjord (Garde et al. 1999) (Fig. 1). Isolated occurrences of low metamorphic grade conglomerates, feldspathic and quartz arenites assumed to have been deposited unconformably on an Archaean basement are exposed on nunataks west and NW of the peninsula Puisortoq (Fig. 1). South of Puisortoq, on either side of Tunua sound (Fig. 1), a 5 km thick supracrustal sequence metamorphosed to upper amphibolite facies is present. Psammites, meta-andesites, minor metamorphosed conglomerates, siltstones, mudstones, calcareous rocks and thin horizons of amphibole- and diopside-bearing intermediate metavolcanic rocks are tentatively correlated with the basal part of the Vallen Group of the NWBZ. Equivalents of the Sortis Group basic metavolcanic rocks are absent. The Ketilidian orogen - major Garde et al. (20020) reported a preliminary ion probe study of zircons from supracrustal rocks of components the NEBZ. The results showed a complex distriThe core of the orogen (Fig. 1) is a 100-200 km bution with a maximum age of c. 1850 Ma resultwide, NE-trending continental magmatic arc, the ing from (i) detrital zircon cores derived from the Julianehab batholith (JB), constructed by juvenile Julianehab batholith, (ii) recrystallization and calc-alkaline magmas emplaced during north- overgrowth at c. 1800-1780 Ma and (iii) variable directed oblique subduction of an oceanic plate lead loss during a subsequent heating episode. (Chadwick & Garde 1996; Garde et al. 20020). The Julianehab batholith is separated from the Archaean foreland by the c. 50 km wide Border Julianehab batholith Zone exposed in the northwestern and northeastern The Julianehab batholith, exposed over a parts of the orogen; however, these two regions are c. 30 000 km2 wedge-shaped area of south Greenseparated by the Greenland Ice Cap (Fig. 1). To the land (Fig. 1), was emplaced between 1850 and SE beyond the outboard margin of the batholith, 1800 Ma (Hamilton et al. 1996; Hamilton 1997; the Psammite and Pelite Zones have been collec- Garde et al. 20020) and is clearly younger than tively interpreted as the relics of the fore-arc the Border Zone supracrustal sequence that it basin (Chadwick & Garde 1996; Garde et al. intrudes. Intrusive ages for the entire batholith 2002a). A prominent late- to post-orogenic suite cluster into two groups centred on c. 1840 and of rapakivi granites sensu Into and related noritic c. 1810 Ma (Garde et al. 20020). Granodiorite plutons are also present in this part of the orogen. and granite sensu stricto predominate along with quartz monzodiorite, tonalite and quartz syenite. Dioritic and more basic lithologies comprise at Border Zone least 5% of the batholith, and form kilometreThe Border Zone (Fig. 1) is defined as the scale distinct intrusions, numerous swarms of synregion of reworked Archaean basement, plutonic mafic dykes and centimetre- to decideformed and recrystallized Palaeoproterozoic metre-sized, fine- to medium-grained dioritic basic Iggavik dykes (c. 2130 Ma) and the enclaves in leucocratic plutons (Chadwick et al. overlying Palaeoproterozoic supracrustal rocks 1994; Chadwick & Garde 1996). Biotite and horn-

Fig. 1. Geological map of the Ketilidian orogen in south Greenland (modified from Garde et al 2002(3). The inset map of Greenland shows the location of the Ketilidian orogen (Ket), the Archaean block of southern Greenland (AB), and the Ammassalik (Am) and Nagssugtoqidian (Nag) orogenic belts. Locations of cross-sections A-A', B-B', C-C' and D-D'-D" are indicated on main map. A, Archaean; S, supracrustal rocks exposed in nunataks.

234

MCCAFFREY ET AL

blende are the main ferromagnesian minerals, but ortho- and clinopyroxene and locally olivine occur in the mafic intrusions. The batholith has a distinct calc-alkaline geochemical character (strong negative Nb, Ta, P and Ti anomalies), rare-earth element (REE) patterns show distinct LREE enrichment with small positive or negative Eu anomalies, and Sr, Pb and Nd isotopic data indicate its juvenile nature (Garde et al 20020). Psammite and Pelite Zones The Psammite Zone and Pelite Zone occupy a minimum width of c. 100 km and comprise deformed and metamorphosed feldspathic arenites (psammites), siltstones and mudstones (pelites), minor limestones and volcanic rocks with an estimated pre-erosional thickness of over 15 km (Garde et al. 20020). Psammitic rocks predominate in the NW close to the batholith (the Psammite Zone), whereas pelitic rocks are widespread in the SE (the Pelite Zone). An excellently preserved volcano-sedimentary sequence was deposited unconformably on the Julianehab batholith at Kangerluluk (Mueller et al. 2000). The Psammite Zone around N0rrearm also contains broadly concordant sheets of hornblende granodiorite, (e.g. sample 412056 dated at 1792 + 1 Ma by Hamilton 1997), diorite, gabbro and gabbro-anorthosite, which are lithologically indistinguishable from intrusive suites in the Julianehab batholith (Garde et al. 20020). The largest of these, the Stendalen complex west of N0rrearm (Fig. 1), is a c. 8 km wide and minimum 1300 m thick, saucer-shaped layered intrusion of gabbro, leucogabbro and anorthosite. Both the Psammite and Pelite Zones have been intensely deformed and have undergone metamorphism to migmatite grade (Garde et al. 20020). In the Pelite Zone, migmatized paragneiss predominates south of Lindenow Fjord; however, anatectic garnetiferous granites and components of the rapakivi granite suite underlie much of the region from Kap Ivar Huitfeldt to Kap Farvel (Fig. 1). Intrusions of the rapakivi suite collectively cover c. 3000km2 and were emplaced mainly into supracrustal rocks of the Psammite and Pelite Zones between 1755 and 1723 Ma (Fig. 1; van Breemen et al 1974; Gulson & Krogh 1975; Hamilton 1997; Garde etal 20020). Border Zone structure Border Zone-foreland boundary relationships In the northwestern Border Zone, Palaeoproterozoic reworking of the Archaean margin is shown

by the NE-trending Iggavik dyke swarm of the foreland becoming increasingly reorientated, deformed and recrystallized southwards (Berthelsen & Henriksen 1975) (Fig. 2a). Henriksen (1969) noted that north of Arsuk Brae the dykes are bent and open folded whereas to the south on the Ivittuut peninsula they are boudinaged into trains of amphibolite lenses and almost conformable to the NNW-trending gneissic foliation. The overlying basal unconformity and Ketilidian cover rock sequence also become increasingly modified by ductile deformation southwards over a distance of c. 100 km (Bondesen & Henriksen 1965). To the south of Arsuk Brae the angular nature of the unconformity is absent as the moderately SE-dipping foliation in the supracrustal rocks is parallel to that in the underlying gneissic rocks (Henriksen 1969). On the east coast, a narrow transition zone approximately 10 km wide between flat-lying, low-grade sedimentary rocks, exposed on nunataks to the north, and intensely deformed and migmatized, steeply SE-dipping supracrustal rocks located south of Puisortoq, takes the form of a NE-trending, composite synform (Fig. 2b). Border Zone structural history The large-scale architecture of the NWBZ is controlled by sinistral transpression partitioned into zones of crustal shortening (folds and thrusts) and zones of sinistral strike-slip deformation (Garde et al. 1998). This main deformation phase produced a major basement-cored, WSW-WNEtrending, anticlinal arch in the region to the west of Graenseland (Fig. 2a). A synclinal structure has also been mapped in the cover rocks at Midternaes to the north. To the south of the main arch, at Arsuk 0, an ENE-trending syncline and anticline pair have been mapped. In the area between Graenseland and Kobberminebugt, sinistral shear produced prominent ENE-plunging, S-shaped folds, which are parasitic on the southern limb of the large anticlinal arch (Fig. 2a). These formed synchronously with intrusion of the Qornoq augen granite (1848 + 2 Ma, Garde et al. 20020). At Kobberminebugt, the emplacement of the Borg Havn granite at 1845±lMa (Garde et al. 20020) post-dated an early fabric and was synchronous with sinistral non-coaxial deformation. Sinistral deformation within the Kobberminebugt shear zone was locally extremely intense and produced steeply plunging, tight folds and refolds of earlier fabrics, sinistral S-C, C' structures and asymmetric boudins. The sinistral transpressional regime was succeeded by a phase of extremely localized dextral non-coaxial ductile deformation dated by the intrusion of the 1805+ 2 Ma, Satukujoq

ATTACHMENT FORMATION DURING PARTITIONING

235

Fig. 2. Geological maps of Border Zone (modified from Garde et al. 1999, 2002a). (a) Northwestern Border Zone. B, Borgs Havn granite; S, Satukujoq granite, (b) Northeastern Border Zone. Inset map shows location of more detailed maps.

granite at Kobberminebugt (McCaffrey et al 1998). Structures that predate the main sinistral transpression have been recognized locally in the Border Zone. There are north- and NNE-trending upright, gently plunging folds and NW-directed thrusts in Midternaes and Graenseland and a prominent top-to-NW transport thrust with contemporaneous tight overturned folds developed in the underlying and intercalated sedimentary rocks of the Sortis Group. The earliest recognized structures preserved in low-strain windows in the Kobberminebugt shear zone are steeply plunging folds with dextral shears on fold limbs (McCaffrey et al. 1998) (Fig. 2a). From these observations, we suggest that an early (pre-1850 Ma) phase of top-to-NW thrust-

ing and margin-parallel dextral shear predated the main sinistral transpression. In the NEBZ, migmatitic feldspathic psammites and deformed plutonic rocks related to the Julianehab batholith generally display intense, steeply SE-dipping LS fabrics with a margin-parallel (NE - SW-trending) stretching lineation and kinematic indicators that indicate sinistral shear. In western Napasorsuaq Fjord, a minimum age for the supracrustal rocks and the intense sinistral deformation is provided by a discordant, but folded, 30 cm thick granite dyke at 1800+1 Ma (Garde et al 2002^). Late, kilometre-scale NE-SW- to ENE-WSWtrending open, upright or steeply north-inclined antiforms and synforms fold this main fabric. To the east of Tunua Sound the southeastern

236

MCCAFFREY ETAL

limb of the late NE-trending, steeply inclined synform (Fig. 2b) displays top-to-NE kinematics. This style of deformation is similar to the main deformation observed in the batholith and forearc to the south on the east coast (see below). To summarize, the Border Zone experienced a complex tectonic history with (i) pre-1850Ma thrusting and dextral shear, (ii) c. 1845 Ma sinistral shear and plutonism (NWBZ), (iii) c. 18001780 Ma sinistral and top-to-NE shear coeval with plutonism, and high-grade metamorphism (NEBZ), and (iv) local dextral shear and plutonism (NWBZ) at c. 1800 Ma. The available evidence suggests either that the main sinistral transpressional deformation migrated away from the northwestern Border Zone towards the east or that there were two episodes of sinistral deformation, one at 1845 Ma and one at c.l 800 Ma. Julianehab batholith structure Border Zone-Julianehab batholith boundary The boundary between the Border Zone and the Julianehab batholith to its south is defined on the west coast by the ENE-trending, vertical Kobberminebugt shear zone (Fig. 2a), which comprises syn- and post-kinematic batholithrelated granitic plutons intruded into upper amphibolite facies supracrustal rocks (Watterson 1968). On the east coast, the equivalent boundary has been mapped along Napasorsuaq Fjord (Fig. 2b), where it represents the southernmost exposures of the supracrustal rocks that have been intruded by Ketilidian plutonic rocks. Here the boundary is a partially reworked intrusive contact. Julianehab batholith, west coast structure Despite being lithologically similar throughout, the Julianehab batholith displays important differences in structural geometry between the west and east coasts. Most of the western parts of the Julianehab batholith were emplaced between c. 1850 and 1800 Ma ago synchronously with deformation in an ENE-WNW-trending sinistral transpressive tectonic setting (Chadwick et al 1994; Chadwick & Garde 1996). South of Qaqortoq (Fig. 1), granitic, intermediate to mafic dykes were emplaced into a system of steep to vertical NNE- to NE-trending, sinistral and subordinate ESE-trending dextral synplutonic shear zones (Watterson 1968; Chadwick et al 1994; Chadwick & Garde 1996). Detailed contact relationships for larger plutons are

unknown except in the NW part of the batholith near Nunarrsuit (Fig. 1) where Pulvertaft (1977) has mapped steep NE-SW-trending contacts between kilometre-sized granitic bodies. A steep NE-SW-trending fabric is present throughout most of the batholith on the west coast. Mineral textures, such as the parallel alignment of subhedral, oscillatory zoned plagioclase phenocrysts set in a finer-grained matrix of quartz and feldspar with undulatory extinction indicate magmatic state deformation (Chadwick et al. 1994). These magmatic state fabrics are overprinted by coplanar LS fabrics with subhorizontal stretching lineations and sinistral S-C fabrics indicative of solid-state strain. Solid-state strain intensifies towards kilometre-scale shear zones, which are mylonitic to ultramylonitic in their central parts. The largest of these is the 1.5km wide Sardloq shear zone (Fig. 1) in the southwestern part of the batholith, which has been dated as being active from 1810 to 1820 Ma (Garde et al 20020). East coast batholith structure In contrast to the steep magmatic and tectonic fabrics that predominate in the northwestern and southwestern parts of the batholith, intrusions over large parts of the batholith in its eastern part were emplaced with moderately inclined contacts (Garde et al. 1999, 2002a). These flat-lying or gently inclined contacts along with a locally intensely developed solidstate fabric have been folded by kilometre-scale late-stage folds into moderately to steeply NWor SE-dipping panels. Unfolding the late folds restores intrusive contacts and magmatic fabrics to subhorizontal (see below). Swarms of gently inclined, undeformed intermediate to mafic silllike sheets are also present in this part of the batholith. The main solid-state foliations throughout the batholith on the east coast are defined by crystallographic and shape preferred orientations of ferromagnesian minerals and plagioclase. These fabrics are moderately to steeply dipping due to the effect of later folding (see below) but generally strike NE-SW, displaying lineations that have moderate to shallow plunges. The main kinematic indicators in the granitic rocks are cr-shaped plagioclase porphyroclasts and a crystallographic preferred orientation in quartz. These show that the earliest pervasive solidstate fabrics were mostly consistent with a transport direction of top to the NE when the effects of later upright folding are removed. At Igutsaat Fjord and Kangerluluk in the SE (Fig. 3), there is clear evidence for at least two

ATTACHMENT FORMATION DURING PARTITIONING

Fig. 3. Geological map of eastern exposures of the Ketilidian orogen showing the positions of late fold and shear zone structures. The boundaries between the Border Zone, Julianehab batholith, Psammite and Pelite Zones are indicated (long thin dashes).

237

238

MCCAFFREY ET AL

phases of pluton intrusion and solid-state deformation. This has resulted in early granodiorite orthogneiss fabrics (protoliths dated at c. 1845 Ma) being intruded by gently inclined, granitic sheets at c. 1794 Ma (Garde et al. 2002a). The late granitic sheets display geometrically similar but less intensely developed magmatic and solid-state fabrics. A series of kilometre-scale, ENE-trending open fold structures have reoriented the earlier magmatic solid-state fabrics into NW- and SEdipping panels (Fig. 3). These antiforms and synforms are spatially associated with zones of intense solid-state strain in a number of NEtrending high-strain zones. They are commonly cut by thin granite sheets or dykes indicating that the high-strain zones were developed before the magmatic activity had finally ceased. Stretching lineations within the latest highstrain zones have oblique plunges, which combined with reverse slip indicators show that these shear zones accommodated an additional component of NW-SE-directed crustal shortening. The late-tectonic Anorituup shear zone (Fig. 3) is a 5 km wide east-west-trending vertical shear zone that displays abundant kinematic indicators, e.g.
Psammite and Pelite Zones Julianehdb batholith-Psammite Zone boundary The unconformable nature of the Julianehab batholit-Psammite Zone boundary is confirmed by the age and provenance of the Psammite Zone sedimentary infill (Garde et al. 2002a). The identification of batholith-derived clasts and U-Pb ages of detrital zircons in the psammites and pelites show that the sediments were sourced primarily from the Julianehab batholith (Hamilton et al 1996; Garde et al. 2002<2). A few Archaean and early Proterozoic zircons are present; however, detrital ages cluster in the 1850 to 1800 Ma range, an exact overlap with the batholith. Furthermore, volcano-sedimentary rocks at Kangerluluk were unconformably deposited on the underlying Julianehab batholith.

The ages of the youngest dated detrital zircons point to c. 1790 Ma as the maximum age for the uplift of the fore-arc basin (Garde et al. 2002<2). The boundary, corresponding to the arc-forearc contact, may be traced from Nanortalik on the west coast to Kangerluluk on the east coast (Fig. 1). The westernmost part of the boundary between the fore-arc and the batholith, near Nanortalik, is obscured by younger granites, but data from largely inaccessible outcrops suggest that it is intrusive or a tectonically modified unconformity (Garde et al. 2002(2). Further to the NE, the unexposed boundary runs along S0ndre Sermilik Fjord and is most likely to be a steep fault or shear zone. At Ippatit, the boundary diverts to an east-west orientation, with the Psammite Zone rocks structurally above the Julianehab batholith (Fig. 4a) and is a modified original unconformity. At N0rrearm, the Julianehab batholith has been thrust southeastwards over the Psammite Zone (Garde et al 2002(2) (Fig. 4b). In inner Danell Fjord (Fig. 4c) the contact between the batholith and the fore-arc is a steep, NE-trending sinistral ductile shear zone and finally becomes part of an east-west shear zone south of Igutsaat Fjord (Fig. 3). Psammite and Pelite Zone structure The structure and tectonic setting of the Psammite Zone has been described in detail by Garde et al (2002&) and therefore is only summarized briefly here. The extension of this tectonic framework into the Pelite Zone was addressed by Chadwick et al (2000) and is also outlined here. Metamorphic and timing constraints on the fore-arc tectonic development are reported by Garde et al (2002(2). Four deformation phases, for convenience named D t to D4, and one localized phase, D5 have been recognized in the Ketilidian fore-arc. The overall structure is generally that of flatlying structures comprising intense composite LS fabrics and tight to isoclinal folds (Fig. 4) (Garde et al 2002&). Higher structural levels indicated by lower metamorphic grades are preserved close to the batholith (Garde et al 2002a, b). Graded greywackes east of the sheared Julianehab batholith contact are folded into tight, upright to SE overturned, gently ENE-plunging F! folds with a spaced axial planar Si cleavage (Fig. 4c). Changes in cleavage and minor fold vergence indicate the presence of a major F! structure close to the batholith at Inner Danell Fjord (Fig. 1). Elsewhere in the Psammite Zone, major tracts of overturned metasedimentary rocks indicate early fold structures;

Fig. 4. Sections to illustrate the main structure of the Psammite Zone, (a) Cross-section A-A' to north and south of Ipatit Valley (from McCaffrey, unpublished data), (b) Cross-section B-B' from inner N0rrearm to inner Kangerluaraq. (c) Cross-section C—C' from 'Sorte Nunatak' to central Danell Fjord, (d) Cross-section D-D'-D" from Stendalen to central Lindenow Fjord. See Figure 1 for locations. Diagrams (b)-(d) modified from Garde et al. (20020, b).

240

MCCAFFREY ETAL.

however, incomplete exposure and intense subsequent deformation make the early geometries indeterminable. From the batholith margin southeastwards, the metamorphic grade and intensity of D2 deformation increase. Early fabrics have been transposed to a composite Si/S2 LS fabric with a marked NE-trending extension lineation, and F2 folds deform D! neosome seams. Migmatite seams along F2 axial surfaces show that anatexis was synchronous with D2. Steep Di~D2 structures in inner Danell Fjord become more shallowly dipping eastwards around the hinge of a large F3 recumbent fold (Fig. 4c). Further SE, the intense Di-D 2 fabric becomes gradually overprinted by recrystallization and extensive partial melting, which yielded cordieritegarnet-bearing neosome and parautochthonous masses of heterogeneous peraluminous granite with metre-sized pockets of refractory siliceous and calcareous rocks. Chadwick et al (2000) established that the main Di-D 2 fabric and general flat-lying structure with overturned to recumbent D3 folds of the Psammite Zone persist southwards into the Pelite Zone and are equivalent to the main fabric of the 'flat-lying migmatite' region described by Bridgwater et al. (1966). The D!-D3 structures were accompanied by anatectic peraluminous granite intruded as major (kilometre-scale) flat-lying parautochthonous intrusive sheets in the Prins Christian Sund area (Fig. 1). This was accompanied by the emplacement of batholith-related felsic to mafic intrusions and thin, but numerous, intermediate to ultrabasic dykes and sills (appinite). In the southwestern Psammite Zone (Escher 1966; Dawes 1970) the main fabric is a composite Di-D 2 structure with a generally flat-lying orientation similar to that east of N0rrearm (Fig. 4b). At Ippatit (Figs 1 & 4a), the Quinqassat shear zone is a D2 NNE-vergent thrust structure as indicated by asymmetric boudins, folds and minor shear planes. Common kinematic indicators east of N0rrearm such as S-C fabrics and asymmetric vein quartz lenses indicate D2 top-to-NE transport (Garde et al. 2002£) (Fig. 4b). The adjacent Julianehab batholith west of N0rrearm consists of steep to vertical granodiorite and diorite sheets with steep solid-state planar fabrics and localized sinistral shear zones. These structures gradually shallow and become NE-dipping moving ESE (Fig. 4b). A complete absence of new fabrics, such as an S3 cleavage similar to that widely developed in Danell Fjord (Fig. 4c), associated with this D2 overturning suggests that it represents the original orientation of the D2 strain trajectories and bears no relation to D3 or D4 folding.

D3 structures include large-scale close to tight folds overturned to the NW with gently dipping NE-striking axial surfaces and NE- or SW-plunging hinges (Garde et al. 2002£). An associated fabric crenulates L2 and S2 and is present mainly in the short, steep fold limbs. The fourth phase of deformation (D4) is most widely developed in the outer part of the fore-arc and produced kilometre-scale open to tight antiforms and tight synformal cusps, with NE-trending upright axial surfaces and shallow NE plunge (Fig 3 & 4). These folds postdate the metamorphic peak (syn-D3) as they fold the main migmatite fabrics. Late folds (F5) with variable, mostly NW, trend appear to be confined to the east coast and are relatively open, large-scale warps and interfere with F4 folds to form open domes and basins. In the Pelite Zone, the generally shallow Di-D 3 fabrics in the migmatites and anatectic granites are gently warped into upright arches and troughs which mainly trend NW-SE (Fig. 3) and are likely to be equivalent to F5 folds further north. This warping appears to be closely related to the emplacement of rapakivi granites (Grocott et al. 1999; Chadwick et al. 2000). Nature and timing of structural and metamorphism events in the fore-arc Garde et al. (2002a) discussed preliminary P-T determinations of garnet-biotite-sillimanite + cordierite-bearing paragenesis in schists of the fore-arc. They obtained peak metamorphic conditions of c. 3 kbar, 580 °C at the batholith margin in inner Danell Fjord, increasing to c. 5 kbar, >800°C south of Prins Christian Sund (Fig. 1). Previously reported lower peak metamorphic pressures of 2-4 kbar at 650800 °C (Dempster et al. 1991) could not be replicated. Isotopic ages of syn-D2 batholith-related intrusions and syn-D3 S-type granites show that D2-D3 deformation and partial melting took place in a short interval, c. 1792-1785 Ma (Fig. 2; Hamilton et al. 1996; Hamilton 1997; Hamilton, unpublished). The rapakivi suite The rapakivi suite occurs mainly as extensive flat-lying intrusive sheets within the fore-arc and along its boundary with the Julianehab batholith (Bridgwater et al. 1974; Grocott et al. 1999). Rapakivi granite exposures between Kap Ivar Huitfeldt and Kap Farvel (Fig. 1) are now considered to be all part of a single 40 km long, more than 2.5 km thick, flat-lying sheet which

ATTACHMENT FORMATION DURING PARTITIONING

is arched over large-scale (F4) antiformal and synformal structures in the supracrustal package (Chadwick et al 2000; Fig. 4). All contacts are sharp and with few exceptions undeformed, and in several places the rapakivi granite sheet is clearly discordant to flat-lying D!-D2 fabrics, D3 migmatite structures, and S-type granites. Intrusion of a vertical NW-SE-trending sheeted dyke complex north of Pins Christian Sund (Fig. 1; Harrison et al 1990) with mingled felsic, mafic and rapakivi granite was synchronous with dextral shear parallel to its wallrocks. Xenoliths of country rock hornfels are common in the basal parts of the rapakivi sheets (Grocott et al. 1999; Chadwick et al 2000). On Qernertoq (Fig. 1), a c. 100 m thick retrograde metamorphic aureole is developed below the contact where the peak metamorphic minerals are partially resorbed, and retrograde cordierite coronas around garnet as well as sheaves of secondary dark red biotite were formed, presumably aided by fluids released during crystallization of the rapakivi granite. The geochronological data provide important additional evidence in support of the field evidence that the main penetrative structures and regional metamorphism are older than emplacement of the rapakivi suite, contrary to the interpretation by Hutton et al (1990) and Dempster^ al (1991). Tectonic evolution of the Ketilidian orogen The most recent model for the tectonic evolution of the Ketilidian orogen (Chadwick & Garde 1996) suggests that the Julianehab batholith formed as the root of a magmatic arc emplaced during sinistral transpression resulting from oblique convergence. Subsequent studies have now largely confirmed and developed this model (Garde et al 2002a). The entire Ketilidian orogenic cycle spans c. 450 Ma but nevertheless the magmatic and tectonic activity appear relatively coherent and no extraneous terranes have been positively identified (Chadwick & Garde 1996; Garde et al 20020). The tectonic evolution of the Ketilidian orogen has been subdivided into five tectonic episodes on the basis of kinematics/ geometry, age or geographical extent of the major structural elements and is summarized below and in Figures 5 and 6: Episode I. Rifting of the Archaean margin followed by deposition of the supracrustal cover sequence in the Border Zone (~2130 to > 1848 Ma).

241

Episode 2. Thrusting of supracrustal sequence over basement in northwestern Border Zone (local DO correlated with dextral strike-slip at the margin in Kobberminebugt (> 1848 Ma). Episode 3. Early sinistral transpression recognized in northwestern Border Zone (local D2) at 1848 to > 1825 Ma along with intrusion of early Julianehab batholith components. Episode 4. Main partitioned sinistral transpression (c. 1825-1785 Ma) forms bulk of Julianehab batholith with mainly steep fabrics and subhorizontal lineations. In the fore-arc, D{ steep fabrics and shallow lineations overprinted by D2 top-to-NE transport on gently inclined structures, then folded by overturned D3 structures. Coeval partitioned top-to-NE deformation and sinistral shear occurred in the northeastern Border Zone. Episode 5. Late sinistral transpression forms ENE-trending sinistral/oblique shears and east-west- and NW-SE-trending dextral shears and folding (F4 and F5 in fore-arc). Deformation coupled with emplacement of rapakivi granites c. 1750-1730 Ma and predominates in the fore-arc and eastern parts of the Julianehab batholith. The earlier Ketilidian episodes (1-3 above) have correlatives in the Makkovik orogen, eastern Labrador, which is the nearest Palaeoproterozoic orogen located to the west along the same Archaean margin (Kerr et al 1997; Culshaw et al 2000; Ketchum et al 2002; Garde et al 20020). Episodes 4 and 5 are more difficult to correlate, an aspect that is discussed later. Contrasting structural histories in the Border Zone The principal preserved structural geometries reflect which of the tectonic episodes have exerted most influence over particular parts of the orogen (Fig. 5). The northwestern Border Zone has been influenced by episodes 1-3 and produced the foreland dyke swarms, the unconformable cover sequence and the early thrusts and dextral shears. However, the largest and most profoundly developed structures relate to the early phase of sinistral transpression (episode 3). This formed the ENE-trending, gently plunging basement-cored arch folds and the Kobberminebugt sinistral shear zone (Fig. 6a). In contrast, the northeastern Border Zone structure appears to mainly reflect episodes 4 (main phase sinistral transpression) and 5 (late folding and thrusting) and the earlier history may not have been developed.

242

MCCAFFREY ET AL.

Fig. 5. Spatial and temporal distributions of the five recognized tectonothermal episodes in the Ketilidian orogen. (a) Map shows spatial distribution of the five episodes, (b) Temporal relationship between the five episodes and the main zones in which it has been observed. BZ, Border Zone; JB, Julianehab batholith; PsZ/PeZ, Psammite and Pelite Zones; tp, transpression.

ATTACHMENT FORMATION DURING PARTITIONING

243

Fig. 6. Possible evolution of the Ketilidian orogen. The structural development of the Border Zone, syntectonic intrusion of the Julianehab batholith, development and exhumation of the fore-arc attachment structure and intrusion of the late-tectonic rapakivi suite may all be related to a framework of protracted sinistral transpression. JB, Julianehab batholith; KsZ, Kobberminebugt shear zone; Ssz, Sondre Sermilik shear zone.

244

MCCAFFREY ETAL.

Julianehdb batholith construction and deformation The age distribution and the structure of the Julianehab batholith were controlled by both the early and main episodes of sinistral transpression (episodes 3 and 4 above). Two major pulses of intrusion occurred between 1854-1836 Ma and 1818-1799 Ma. No zircon ages have been recorded so far within the interval 1835-1819 Ma (Garde et al 20020). The older phases are preserved in the external northwestern and the southern parts of the batholith (Fig. 5). Pluton contacts mapped of the northwestern batholith (Pulvertaft 1977) are consistent with emplacement during the same early sinistral episode that deformed the foreland and cover rocks. The later plutonic phases appear to have infilled the central batholith. A restoration by matching the spatial patterns of the older phases suggests that the younger pulses were also emplaced during sinistral deformation (Fig. 6b). Fabric geometries within the batholith also reflect sinistral transpression. On the west coast the major structures are steep shear zones with orogen-parallel sinistral transport that formed during the early episode of sinistral transpression at Kobberminebugt and Niaqonaarsuk and the later episode at Sardloq (Fig. 1 for locations). On the east coast, steep sinistral shear zones of the later phase are found at the margins of the batholith (Napassorsuaq, inner Danell Fjord and Paatusoq) (Fig. 1). The intervening batholith is characterized by gently inclined primary structure with orogen-parallel lineation formed during episode 4. The nature of these geometries is discussed further below. This part of the batholith (Napassorsuaq to Kangerluk) has also been affected by episode 5 which produced the NEtrending upright, open folds and steep reverse thrusting.

Psammite and Pelite Zones - development of a mid-crustal attachment structure The structures of the Psammite and Pelite Zones reflect the later episodes (4 and 5) of the Ketilidian orogen. Garde et al (2002£) have developed a model to explain the geometric and kinematic development of the Psammite Zone during the main phase of sinistral transpression (episode 4). In this model, transpression during D! was partitioned into sinistral strike-slip in the magmatic arc rocks of the batholith and overthrusting in the sedimentary and volcaniclastic sequences of the proximal fore-arc. Except for isolated regions near the batholith (e.g. inner Danell

Fjord), D! fabrics in most of the Psammite Zone were overprinted and transposed by the intense, flat-lying, D2 ductile shear zone with arc-parallel stretching lineations and top-to-NE kinematic indicators. Further south, the geometry of the flat-lying migmatites, the grades of metamorphism and formation of large anatectic granite sheets of the Pelite Zone are compatible with the Psammite Zone D2 deformation style, except that it formed at deeper levels (Chadwick et al. 2000). The D2 shear zone is interpreted as lying beneath the domain of strike-slip partitioned transpression in the fore-arc (Fig. 7). Such a shear zone would be required to accommodate movements on the partitioned blocks at higher levels and on the grounds of kinematic compatibility between crustal levels; we have named it an attachment shear zone. Overprinting of the domain of DI strike-slip partitioned transpression that existed at a higher structural level in the fore-arc by the initially deeper-level D2 attachment is attributed to its migration to a higher crustal level (discussed later). The latest significant deformation (episode 5) recorded in the Psammite Zone and Pelite Zone produced the NE-trending F4 and predominately NW-trending (F5) large-scale open warp-like fold structures (Fig. 3). These appear to be coeval with ENE-trending sinistral oblique, east-west-trending dextral and NW-SE-trending dextral shear zones. The overall tectonic regime at this time is consistent with a weak sinistral transpression (Fig. 6c). The F4 and F5 folds interfere to form kilometre-scale domes and basins, and the hinge zones of a number of these coincide with the location of the rapakivi granite intrusions (Chadwick et al. 2000).

Fig. 7. Schematic block diagram of the batholith and adjacent fore-arc structure showing D! partitioned transpression and mid-crustal attachment. Subsequently during D2 and D3 the attachment migrated into the upper crust before being intruded by rapakivi suite (see text for discussion). Modified from Garde et al. (20020, b}.

ATTACHMENT FORMATION DURING PARTITIONING

Rapakivi granite intrusion The emplacement mechanism and geotectonic setting of the rapakivi suite have been thoroughly debated in the literature (Escher 1966; Bridgwater et al 1974; Dempster et al 1991; Windley 1991; Hutton et al 1990, 2000; Grocott et al 1999). We conclude that the regional and detailed structure of the Prins Christian Sund rapakivi granite and other members of the rapakivi suite, their contact relations, U-Pb zircon geochronology, microtextures and P-T determinations of contact metamorphic host rocks show that they were emplaced as giant flat-lying approximately tabular sheets by roof uplift or floor depression, and fed by narrow vertical dykes. We find no evidence to support previous studies that suggested that the fore-arc high-temperature, low-pressure (HT-LP) metamorphism and emplacement of the rapakivi suite took place during post-orogenic extension (Hutton et al 1990; Dempster et al 1991). Our studies show that a contractional (transpressive) tectonic environment persisted throughout the later evolution of the orogen.

Oblique convergence and development of attachment zone in Ketilidian orogen It has been recognized since the work of Fitch (1972) that oblique convergence at continental margins produces partitioning of the deformation into orogen-parallel and orogen-normal components. However, as pointed out by Saint Blanquat et al (1998), strike-slip partitioning may occur for active magmatic arcs with a convergence angle as high as 85°, and thus little tangential plate motion may in fact be necessary. This is due to the arc forming a vertical, lithospheric-scale planar weakness that localizes the strike-slip component of the strain (Beck 1983; Saint Blanquat et al 1998). The model proposed by Garde et al (2002/?) explains the batholith/fore-arc structural geometries as resulting from a partitioned D! sinistral transpression progressively overprinted by (D2) flat-lying planar fabrics with top-to-NE transport. This is attributed to the kinematically necessary underlying attachment zone, initially located in the mid-crust, migrating upwards and overprinting the partitioned fabrics (Fig. 7). We have now established that the flat-lying top-to-NE and steep sinistral deformation also formed in both the Julianehab batholith and the Border Zone exposed in the eastern coastal sections of the Ketilidian orogen (see above). Thus, the c. 1800 Ma attachment zone deformation is widely distributed within the arc and back-arc regions as well as the

245

fore-arc-batholith boundary as detailed by Garde et al (2002Z?). If the arc acted as a vertical zone of weakness, then horizontal transmission of transpression into the back-arc would have been limited. In this scenario, the observation of attachment-style deformation is important as it shows that an upward mechanical coupling must have occurred from the lower crust/upper mantle to the upper crust in order for the partitioned strike-slip deformation to be transmitted into the back-arc region as well as the fore-arc. This type of basal forcing has been proposed for the San Andreas fault (Teyssier & Tikoff 1998) and other magmatic arc systems (Oldow et al 1990; Saint Blanquat et al 1998). The kinematics and geometry of structures in the attachment zone are consistent with subhorizontal arc-parallel (NE) transport of an overlying (now mostly removed) fore-arc sliver along the continental margin. Our studies of the nature of the attachment zone that developed in the Ketilidian orogen have revealed a number of significant aspects. Firstly, our tightly constrained structural history indicates that the main fore-arc tectonothermal evolution occurred within a short (c. 10-15 Ma) time-span relative to the protracted sinistral transpression that occurred over 120 Ma for the entire orogen. The inference from this is that this was the length of time over which the attachment zone was actively migrating to shallower levels. Secondly, there is a close association between attachment-style deformation and the development of migmatites and larger sheets of garnet/cordierite-bearing peraluminous granite. Anatexis of supracrustal rocks in the back-arc region has also been observed. Thirdly, the attachment had undergone cooling and/or uplift above the brittle-ductile transition by the time of the intrusion of the rapakivi suite some 30-50 Ma after its deformation and metamorphic peak. We speculate that the short time duration of the fore-arc deformation and metamorphism may have been due to plate-scale processes, e.g. the end of oblique subduction (due to subduction choke?) followed by decoupling of the downgoing slab would allow an upwelling of hot asthenosphere to impact the overriding plate. The enhanced geothermal gradient calculated for the fore-arc and the nature of the HTLP metamorphism supports this hypothesis. Mantle-derived magmas such as the Stendalen gabbroic complex and sheets of diorite and arc-granites, which intruded the supracrustal pile during the Di-D 2 deformation, are likely to have contributed some of the heat. Mafic underplating has also been suggested as a

246

MCCAFFREY ET AL.

possible heat source (Dempster et al. 1991; Chadwick & Garde 1996) and numerous intermediate to ultrabasic (appinite) dykes are present throughout the entire fore-arc. The short duration of the deformation and metamorphism in the fore-arc may have been contributed to by a rapid exhumation of the fore-arc during ongoing plate convergence (Willett et al. 1993; Jameson et al 2002). Widespread partial melting and anatectic granite intrusion associated with the development of the Ketilidian attachment structure suggests that there was likely to have been a significant weak layer in the midcrust. Even small percentages of partial melt have a profound weakening effect in the midcrust (Handy et al. 2001) and in convergent orogens may lead to the preferential exhumation of the fore-arc by erosion (Jameson et al. 2002). During or shortly after exhumation, the migmatites and the anatectic peraluminous granites would have crystallized, resulting in strengthening of the attachment zone as it moved upwards. The final episode of mild transpression was synchronous with rapakivi granite intrusion and characterized by large-scale (10km) wavelength buckle folds, indicating that, by c. 1750 Ma, a switch in deformation mode to brittle behaviour typical of the upper crust had occurred in the fore-arc. Patterns of deformation and metamorphism in the orogen show a general outboard migration with time during prolonged sinistral transpression (Figs 5 & 6). Within this framework, the strike-slip partitioning and the thermal history in the Ketilidian fore-arc produced deformation that was highly focused in time and space, and was responsible for producing the observed differences in the structural history and geometry between the east and west coasts in both the Border Zone and Julianehab batholith. External correlation to other nearby orogenic belts (e.g. Makkovik orogen) (Garde etal 2002a; Ketchum etal 2002) is therefore also likely to be problematic for the same reasons. This is especially true for the outboard parts of orogenic belts where focusing of tectonothermal processes during ongoing convergence may be most marked. Conclusions A mid-crustal attachment zone is exposed in the Palaeoproterozoic Ketilidian orogen, south Greenland, and formed as a result of northdirected oblique convergence at a cordillerantype margin. Rifting at or prior to 2.1 Ga and contractional dextral deformation and magmatism (> 1850 Ma) on the continental margin were followed by an extended period of sinistral

transpression from 1850 to 1730 Ma, now separated into three episodes or peaks of activity. The first sinistral episode was focused on the backarc/foreland region and was followed by the main arc construction phase during which transpression was partitioned into strike-slip and contraction components. Despite the longevity of the sinistral transpression, individual tectonic events took place rapidly, e.g. development of fore-arc D!-D3 and accompanying HT-LP metamorphism took place over c. 12 Ma. We suggest that an important part of the fore-arc/batholith evolution was produced by the upward migration of an underlying attachment structure into the upper partitioned crust. This migration may be attributed to an increase in the geothermal gradient folio wed/accompanied by exhumation of the mid-crust. The partially molten, hence weak, attachment zone solidified and strengthened during and after uplift before emplacement of the post-orogenic rapakivi suite during the last phase of mild sinistral transpression. This paper is published with permission of the Geological Survey of Denmark and Greenland. Support from the Natural Environment Research Council (Research Grant GR3/12070) to KJWMcC and JG and from the Danish Natural Science Research Council and the Carlsberg Foundation to AAG for fieldwork and geochronological studies is gratefully acknowledged. We also thank the staff at Telestation Pins Christian Sund for their hospitality, and helicopter pilots F. Bisgaard, C. Bratt, H. Hammer, S. Forstedt, U. Stoller and P. Westlund for their skilful flying. The authors are also grateful to J. van Gool and T. Brewer for comments on the earlier version.

References BECK, M.E. 1983. On the mechanism of tectonic transport in zones of oblique subduction. Tectonophysics, 93, 1-11. BERTHELSEN, A. & HENRIKSEN, N. 1975. Geological Map of Greenland, 1:100000, Ivigtut 61 V. 1 syd. Descriptive text. Geological Survey of Greenland, Copenhagen. BONDESEN, E. & HENRIKSEN, N. 1965. On Some PreCambrian Metadolerites from the Central Ivigtut Region, SW Greenland. Bulletin Gr0nlands Geologiske Unders0gelse, 52. BRIDGWATER, D., SUTTON, J. & WATTERSON, J. 1966. The Precambrian rapakivi suite and surrounding gneisses of the Kap Farvel area, South Greenland. Rapport Gr0nlands Geologiske Unders0gelse, 11, 52-54. BRIDGWATER, D., SUTTON, J. & WATTERSON, J. 1974. Crustal downfolding associated with igneous activity. Tectonophysics, 21, 51-11. CHADWICK, B. & GARDE, A.A. 1996. Palaeoproterozoic oblique convergence in South Greenland: a reappraisal of the Ketilidian Orogen. In: BREWER, T.S. (ed.) Precambrian Crustal Evolution in the North Atlantic

ATTACHMENT FORMATION DURING PARTITIONING Region. Geological Society, London, Special Publications, 112, 179-196. CHADWICK, B., ERFURT, P., etal 1994. Sinistral transpression and hydrothermal activity during emplacement of the Early Proterozoic Julianehab batholith, Ketilidian orogenic belt, South Greenland. Rapport Gr0nlands Geologiske Unders0gelse, 163, 5-22. CHADWICK, B., GARDE, A.A., GROCOTT, J. MCCAFFREY, K.J.W. & HAMILTON, M.A. 2000. Ketilidian structure and the rapakivi suite between Lindenow Fjord and Kap Farvel, SouthEast Greenland. Geological Survey of Greenland Bulletin, 186, 50-59. CULSHAW, N., KETCHUM, J. & BARR, S. 2000. Structural evolution of the Makkovik Province, Labrador, Canada: tectonic processes during 200 Myr at a Paleoproterozoic active margin. Tectonics, 19,961-977. DA WES, P.R. 1970. The Plutonic History of the Tasiussaq Area, South Greenland, with Special Reference to a High-Grade Gneiss Complex. Bulletin Gr0nlands Geologiske Unders0gelse, 88. DEMPSTER, T.J., HARRISON, T.N., BROWN, P.E. & HUTTON, D.H.W. 1991. Low-pressure granulites from the Ketilidian mobile belt of southern Greenland. Journal of Petrology, 32, 979-1004. DEWEY, J.F. 1975. Finite plate evolution: some implications for the evolution of rock masses at plate margins. American Journal of Science, 275A, 260-284. DEWEY, J.F., HOLDSWORTH, R.E. & STRACHAN, R.A. 1998. Transpression and transtension zones. In: HOLDSWORTH, R.E., STRACHAN, R.A. & DEWEY, J.F. (eds) Continental Transpressional and Transtensional Tectonics. Geological Society, London, Special Publications, 135, 1-14. ESCHER, A. 1966. The Deformation and Granitisation of Ketilidian Rocks in the Nanortalik Area. S. Greenland. Bulletin Gr0nlands Geologiske Unders0gelse, 59. FITCH, T.J. 1972. Plate convergence, transcurrent faults, and internal deformation adjacent to southeast Asia and the western Pacific. Journal of Geophysical Research, 77, 4432-4461. GARDE, A.A., CHADWICK, B., MCCAFFREY, K.J.W. & CURTIS, M. 1998. Reassessment of the northwestern border zone of the Palaeoproterozoic Ketilidian orogen, South Greenland. Geology of Greenland Survey Bulletin, 180, 111-118. GARDE, A.A., GROCOTT, J. & MCCAFFREY, K.J.W. 1999. New insights on the north-eastern part of the Ketilidian orogen in South-East Greenland. Geology of Greenland Survey Bulletin, 183, 23-33. GARDE, A.A., HAMILTON, M.A., CHADWICK, B., GROCOTT, J. & MCCAFFREY, K.J.W. 2002a. The Ketilidian orogen of South Greenland: geochronology, tectonics, magmatism and forearc accretion during Palaeoproterozoic oblique convergence. Canadian Journal of Earth Sciences, 39, 765-793. GARDE, A.A., CHADWICK, B., GROCOTT, J. MCCAFFREY, K.J.W. & SWAGER, C.P. 2002£. Mid-crustal detachment during oblique convergence: implications of partitioned transpression in

247

the magmatic arc and proximal forearc of the Palaeoproterozoic Ketilidian orogen, southern Greenland. Journal of the Geological Society, London, 159, 247-261. GROCOTT, J., GARDE, A.A., CHADWICK, B., CRUDEN, A.R. & SWAGER, C.P. 1999. Emplacement of rapakivi granite and syenite by floor depression and roof uplift in the Palaeoproterozoic Ketilidian orogen, South Greenland. Journal of the Geological Society, London, 156, 15-24. GULSON, B.L. & KROGH, T.E. 1975. Evidence of multiple intrusion, possible resetting of U-Pb ages, and new crystallization of zircons in the post-tectonic intrusions ('Rapakivi granites'), and gneisses from South Greenland. Geochimica et CosmochimicaActa, 39, 65-82. HAMILTON, M.A. 1997. Synopsis of new U-Pb geochronological results from the Ketilidian orogen, South Greenland. In: WARDLE, R.J. & HALI, J. (compilers) Lithoprobe Eastern Canadian Shield Onshore-Offshore Transect (ECSOOT): Report of 1997 Transect Meeting. Lithoprobe Report, 61, 118-134. HAMILTON, M.A., CHADWICK, B., GARDE, A.A. & SWAGER, C. 1996. Observations on Palaeoproterozoic fore-arc sedimentation and deformation: preliminary U-Pb results from the Ketilidian orogen, South Greenland. In: WARDLE R.J., & HALL, J. (compilers) Lithoprobe Eastern Canadian Shield Onshore-Offshore Transect (ECSOOT): Report of 1996 Transect Meeting. Lithoprobe Report, 57, 112-122. HANDY, M.R., MULCH, A., ROSENAU, M. & ROSENBERG, C.L. 2001. The role of fault zones and melts as agents of weakening, hardening and differentiation of the continental crust: a synthesis. In: HOLDSWORTH, R.E., STRACHAN, R.A., MAGLOUGHIN, J.F. & KNIPE, R.J. (eds) The Nature and Tectonic Significance of Fault Zone Weakening. Geological Society, London, Special Publications, 186, 305-332. HARRISON, T.N., REAVY, R.J., FINCH, A.A. & BROWN, P.E. 1990. Coexisting mafic and felsic magmas in the early Proterozoic rapakivi suite of Southern Greenland. Bulletin of the Geological Society of Denmark, 38, 53-58. HENRIKSEN, N. 1969. Boundary relations between Precambrian fold belts in the Ivigtut area, Southwest Greenland. Geological Association of Canada Special Paper, 5, 143-154. HUTTON, D.H.W., DEMPSTER, T.J., BROWN, P.E. & BECKER, S.D. 1990. A new mechanism of granite emplacement: intrusion in active extensional shear zones. Nature, 343, 452-455. HUTTON, D.H.W., BROWN, P.E., GROCOTT, J., GARDE, A.A., CHADWICK, B., CRUDEN, A.R. & SWAGER, C.P. 2000. Discussion on emplacement of rapakivi granite and syenite by floor depression and roof uplift in the Palaeoproterozoic Ketilidian orogen, South Greenland. Journal of the Geological Society, London, 157, 701-704. JAMESON, R.A., BEAUMONT, C., NGUYEN, M.H. & LEE, B. 2002. Interaction of metamorphism, deformation and exhumation in large convergent

248

MCCAFFREY ETAL

orogens. Journal of Metamorphic Geology, 20, 924. KERR, A.J., HALL, J., WARDLE, R.I., GOWER, C.F. & RYAN, B. 1997. New reflections on the structure and evolution of the Makkovikian-Ketilidian orogen in Labrador and southern Greenland. Tectonics, 16, 942-965. KETCHUM, J.W.F., CULSHAW, N.G. & BARR, S.M. 2002. Anatomy and erogenic history of a Palaeoproterozoic accretionary belt, the Makkovik Province, Labrador, Canada. Canadian Journal of Earth Sciences, 39, 711-730.

RICHARD, P. & COBBOLD, P. 1990. Experimental insights into partitioning of fault motions in continental convergent wrench zones. Annales Tectonicae, IV, 35-44. ROYDEN, L. 1996. Coupling and decoupling of crust and mantle in convergent orogens: implications for strain partitioning in the crust. Journal of Geophysical Research, 101, 17679-17705. SAINT BLANQUAT, M. (DE), TIKOFF, B., TEYSSIER, C. & VIGNERESSE, J.L. 1998. Transpressional kinematics and magmatic arcs. In: HOLDSWORTH, R.E., STRACHAN, R.A. & DEWEY, J.F. (eds) Continental Transpressional and Transtensional TecKOHLSTEDT, D.L., EVANS, B. & MACKWELL, S.J. tonics. Geological Society, London, Special 1995. Strength of the lithosphere: constraints Publications, 135, 327-340. imposed by laboratory experiments. Journal of STRACHAN, R.A., HOLDSWORTH, R.E., FRIDERICHSEN, Geophysical Research, 100, 17587-17602. J.D. & JEPSEN, H.F. 1992. Regional Caledonian MCKENZIE, D.P. & JACKSON, J. 1983. The relationship structure within an oblique convergence zone, between strain rates, crustal thickening, palaeoDronning Louise Land, N.E. Greenland. Journal magnetism, finite strain and fault movements of the Geological Society, London, 149, 359-371. within a deformation zone. Earth and Planetary TEYSSIER, C. & TIKOFF, B. 1998. Strike-slip partiScience Letters, 65, 182-202. tioned transpression of the San Andreas fault MCKENZIE, D.P. & JACKSON, J. 2002. Conditions for system: a lithospheric-scale approach. In: HOLDSflow in the crust. Tectonics, 21, (6), 1055, doi: WORTH, R.E., STRACHAN, R.A. & DEWEY, J.F. 10.1029/2002TC001394. (eds) Continental Transpressional and TranstenMCCAFFREY, R. 1992. Oblique plate convergence, sional Tectonics. Geological Society, London, slip vectors, and fore-arc deformation. Journal of Special Publications, 135, 143-158. Geographical Research, 97, 8905-8915. TEYSSIER, C., TIKOFF, B. & WATERS, C. 2002. AttachMCCAFFREY, K., CHADWICK, B., GARDE, A.A., ment between brittle and ductile crust at wrenching HAMILTON, M.A. & CURTIS, M. 1998. Preliminary plate boundaries. In: BERTOTTI, G., SCHULMANN, results from the north-western border zone K. & CLOETINGH, S. (eds) Continental Collision of the Ketilidian orogen, South Greenland. In: and the Tectono-sedimentary Evolution of ForeWARDLE, RJ. & HALL, J. (compilers) Lithoprobe lands. European Geophysical Society Special PubEastern Canadian Shield Onshore—Offshore translication, 1, 93-117. ect (ECSOOT): Report of 1998 Transect Meeting. TIKOFF, B. & TEYSSIER, C. 1994. Strain modeling of Lithoprobe Report, 68, 103-114. displacement-field partitioning in transpressional MOLNAR, P. 1992. Brace-Goetze strength-profiles, orogens. Journal of Structural Geology, 16, the partitioning of strike-slip and thrust faulting at 1575-1588. zones of oblique convergence, and the stress- TIKOFF, B., TEYSSIER, C. & WATERS, C. 2001. Clutch heat flow paradox of the San Andreas fault. In: Tectonics and the Partial Attachment of LithoEVANS, B. & WONG, T.F. (eds) Fault Mechanics spheric Layers. European Union of Geosciences and Transport Properties of Rocks. Academic EUG XI programme, 396. Press, London, 435-459. VAN BREEMEN, O., AFTALION, M. & ALLAART, J. MUELLER, W.U., GARDE, A.A. & STENDAL, H. 2000. 1974. Isotopic and geochronologic studies on Shallow-water, eruption-fed, mafic pyroclastic granites from the Ketilidian Mobile Belt of South deposits along a Paleoproterozoic coastline: KanGreenland. Geological Society of America Bulletin, gerluluk volcano-sedimentary sequence, southeast 85, 403-412. Greenland. Precambrian Research, 101, 163-192. WATTERSON, J. 1968. Plutonic Development of the NORTHUP, C.J. & BURCHFIEL, B.C. 1996. Orogen parIlordleq Area, South Greenland. II: Lateallel transport and vertical partitioning of strain Kinematic Basic Dykes. Bulletin Gr0nlands Geoloduring oblique collision, Efjord, North Norway. giske Unders0gelse, 70. Journal of Structural Geology, 18, 1231-1244. WILLETT, S., BEAUMONT, C. & FULLSACK, P. 1993. OLDOW, J.S., BALLY, A.W. & AVE LALLEMENT, H.G. Mechanical model for the tectonics of doubly 1990. Transpression, orogenic float, and lithovergent compressional orogens. Geology, 21, spheric balance. Geology, 18, 991-994. 371-374. PULVERTAFT, T.C.R. 1977. The Archaean and Lower WINDLEY, B.F. 1991. Early Proterozoic collision tecProterozoic (Ketilidian) Rocks of the 1:100000 tonics, and rapakivi granites as intrusions in an Sheet Area Nunarssuit (60 V.I N), South Greenextensional thrust-thickened crust: the Ketilidian land. Unpublished report, Geological Survey of orogen, South Greenland. Tectonophysics, 195, 1-10. Greenland.

Lower Miocene deformation in the hanging wall of the Internal-External Zone boundary of the Betic Cordillera: deformation at the edges of vertical-axis rotation domains in oblique convergent margins E. FERNANDEZ-FERNANDEZ1, A. JABALOY1 & F. GONZALEZ-LODEIRO1 1 Departamento de Geodindmica, Universidad de Granada, 18071 Granada, Spain (e-mail: [email protected]) Abstract: In the eastern Betic Cordillera, non-cylindrical doubly plunging folds deform the Subbetic Zone (the hanging wall of the Internal-External Zone boundary). Their hinges define an arc from north-south trends in the east to ENE-WSW in the west. These folds began to form during the Early Burdigalian with a NNE-SSW trend. Middle Burdigalian rocks define progressive unconformities in the cores of the synforms, recording the tightening of the folds and an increase in the plunge of the fold hinges. All these folds experienced verticalaxis rotations during the Early Miocene, acquiring the present-day arcuate pattern. During the Middle Burdigalian, the thrusting of the External Zones over the Internal Zones occurred with a top-to-the-SE sense of movement. Upper Burdigalian deposits seal the tectonic windows eroded in these thrust surfaces. This deformational history records a constrictional deformation with area reduction in the Subbetic Zone during the Early Miocene, which can be correlated with the dextral oblique convergence between the Internal and the External Zones.

The geometry and kinematics of the structures developing in the overriding plate of an oblique convergence zone were studied during the 1970s and 1980s in active convergent plate boundaries (e.g. Fitch 1972; Beck 1983). These works mention a partitioning of the deformation in several cases, with a strike-slip zone behind the accretionary wedge accommodating the strike-slip component and a nearly normal convergence at the front of the wedge. McCaffrey (1992) and Yu et al. (1993) further developed these models with their observations in active plate boundaries with oblique convergence, relating the angle between the slip vectors of thrust earthquakes with the normal trench axis. McCaffrey (1992) and Platt (1993, 2000) have presented theoretical models on the mechanics of oblique convergence allowing the calculation of the strike-parallel shear within the wedge. The existence of a domain where the transpression is partitioned between strike-slip faults and fold-and-thrust belts that accommodate the normal shortening suggests the presence of a lower detachment level separating this zone from a footwall with a distributed strain (Teyssier & Tikoff 1998). Models assuming that the fold-and-thust belt accommodate only the normal shortening component of the transpression tend to neglect the existence of vertical-axis rotation; only Platt

(2000) has discussed the theoretical possibility of vertical-axis rotation in the lower rear part of a viscous wedge experiencing oblique convergence. However, large vertical-axis rotations associated with detachments have been documented in the upper crustal levels of areas that underwent oblique convergence, as for example in the Andean orocline in Peru and Chile, and also in the Serrania del Interior in Venezuela (Jones et al 1996; McQuarrie 2002). In the San Andreas fault, the vertical-axis rotation has been attributed to the slip of pre-existent vertical faults that individualize the rotated domains (Teyssier & Tikoff 1998). The Internal-External Zone boundary (IEZB) of the Betic Cordillera may stand as a good example of an area that underwent dextral oblique convergence during the Neogene (Lonergan et al. 1994). Moreover, important vertical-axis rotations have been documented in both sides of the IEZB using palaeomagnetic studies (Osete et al. 1988, 1989; Villalain et al. 1992; Platzman 1992, 1994; Allerton et al. 1993; Platt et al. 1995; Platzman et al. 2000; Calvo et al. 2001). These rotations are developed on thrust units not bounded by vertical faults. In this chapter, we analyse the structure and kinematics of the Subbetics in the eastern sector of the Betic Cordillera, in the region north of

From: GROCOTT, J., MCCAFFREY, K. J. W., TAYLOR, G. & TIKOFF, B. (eds) 2004. Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 249-277. 0305-8719/04/$15 © The Geological Society of London 2004.

250

E. FERNANDEZ-FERNANDEZ ET AL.

Fig. 1. Geological map of the Betic Cordillera. The rectangle marks the location of the study area and the thick line corresponds to the location of the cross-section in Figure 2.

Velez Rubio (Fig. 1), where there are good outcrops of the main fault zone associated with the IEZB. The main focus of this chapter is on the geometry and kinematics of the contractional structures developed over the main thrust surface of the IEZB, in order to establish their relationship with the convergence between the Iberian margin and the Internal Zones, and to propose a scenario for the evolution of these contractional structures. This case study is relevant to the deformation in oblique convergence mechanisms because, in this area, the structures observed show the accommodation of the oblique convergence in the uppermost levels of the crust by means of deformation with area reduction and vertical-axis rotation within the thrust sheet. The area also illustrates how, in a domain where the strain of a transpression is partitioned, part of the strike-slip component is accommodated by the internal deformation within the thrust unit.

Geological setting The Betic Cordillera has an ENE-WSW trend and is located in the south and SE of the

Iberian Peninsula. Fallot (1948) grouped the rocks of this mountain chain into the External Zones and the Internal Zones (Fig. 1). The External Zones are a fold-and-thrust belt, which consist of two main geological units: the Subbetic and Prebetic Zones. Both zones developed from the Mesozoic and Tertiary sediments that were deposited on the southern rifted margin of the Iberian plate at the western end of Tethys (Garcia-Hernandez et al 1980). Southwards of the External Zones are the Flysch Trough Units comprising marine siliciclastic sediments of Cretaceous to Early Miocene (Early Burdigalian) age. The Internal Zones (including the basement of the Alboran Sea) comprise the remains of an older collisional mountain belt formed during convergence between Africa and Eurasia during Late Cretaceous(?) to Tertiary times, originally located by most authors to the west of the present-day position. Throughout most of the Internal Zones there is evidence of high-pressure, low-temperature (HP-LT) metamorphism related to continental convergence. Subsequently, in the Early Miocene, the Internal Zones and the Alboran Sea underwent pervasive ENE-WSW

CONSTRICTIONAL DEFORMATION IN THE IEZB

extension (Gonzalez-Lodeiro et al. 1996), coeval with the shortening in the External Zones' foldand-thrust belt (Kirker & Platt 1998; GalindoZaldivar et al 2000; Crespo-Blanc & Campos 2001). The External Zones crop out north of the Internal Zones in an ENE-WSW band (Fig. 1). Azema et al. (1979) and Garcia-Hernandez et al. (1980) grouped the rocks of this zone into two major palaeogeographical units, according to the different facies of the Jurassic rocks younger than the Pliensbachian. These major units are: the Prebetic Zone (Blumenthal 1927) and the Subbetic Zone (Fallot 1945), the latter separated from the former by the Intermediate Units (Foucault 1960, 1962). In the western Betic Cordillera, there are no outcrops of Prebetic Zone rocks; and there is a Neogene basin, the Guadalquivir Basin, between the Subbetic rocks and the Variscan foreland (Fig. 1). A unit of olistostrome deposits (known as the Guadalquivir Olistostromes) occurs between the Subbetic s and the Guadalquivir Basin (Perconig 1962) (Fig. 1). The Triassic successions of the Prebetic and Subbetic Zones and of the undeformed cover of the Iberian Massif are very similar: mainly terrestrial red bed succession with evaporite and carbonate levels. At the end of the Triassic, a wide, shallow carbonate shelf formed throughout the External Zones (Garcia-Hernandez et al. 1980). The differentiation between the Prebetic and the Subbetic Zones began during the Early Pliensbachian, when the carbonate shelf fragmented during Lower Jurassic rifting of Pangaea (Garcia-Hernandez et al. 1980, 1989). From this moment onward, Mesozoic sedimentation in the Prebetic Zone was characterized by shallow-marine deposits, including several coastal and continental episodes. The subsiding trough, filled with rocks of the Intermediate Units, separated the Prebetic Zone from the Subbetic Zone. Pelagic marine facies are common in the Jurassic rocks of the Subbetic Zone with abundant lateral changes of facies. In the Subbetic Zone of the central Betic Cordillera, the old carbonate shelf of the earliest Jurassic became a series of rifted fault blocks separated by subsiding troughs that have been identified using this change of facies (Garcia-Duenas 1967; Azema et al. 1979). During the Late Jurassic-Early Cretaceous, the region underwent a new rifting stage (Rey 1998; Nieto et al 2001; Vilas et al 2001; Fernandez-Fernandez et al 2003), leading to subsidence of the former highs, as deep as below the CCD (carbonate compensation depth) during the Aptian.

251

The External Zones of the Betic Cordillera, characterized by extension during the Mesozoic, began to be deformed during the Burdigalian in a compressional regime as a consequence of the convergence with the Internal Zones. The age of the beginning of the contractional deformations in the External Zones is well constrained in the Early Miocene (Alonso-Zarza et al 2002). However, the end of the extensional deformations in the same area is not well documented: some authors suggest the Early Cretaceous (Rey 1998), while others show evidence of active normal faults during the Palaeogene (FernandezFernandez et al 2003). Most authors propose that the deeper rocks of the Internal Zones underwent HP-LT metamorphism and associated contractional deformation prior to the Early Miocene (Aquitanian). However, the uppermost units (Malaguide Complex) underwent mainly brittle deformation (Lonergan 1993). Only in the base of the Malaguide Complex have some Alpine metamorphism and ductile deformation been documented (Chalouan & Michard 1990; Lonergan & Platt 1995). During the Early Miocene (Aquitanian), the Internal Zones underwent a stage of dramatic thinning and extension that produced a suite of Neogene sedimentary basins, including the present-day Alboran Sea (Sanz de Galdeano et al 1993; Platt et al 1998; Comas et al 1999). This crustal domain migrated westwards and collided with the South Iberian margin with dextral oblique convergence kinematics (Balanya & GarciaDuenas 1987; Lonergan et al 1994). Garcia-Duenas (1967), Garcia-Hernandez et al (1980), Frizon de Lamotte et al (1991) and Galindo-Zaldivar et al (2000) have studied the main features of the structure of the External Zones in the central sector of the cordillera. Moreover, Paquet (1969), Garcia-Hernandez et al (1980), de Smet (1984), de Ruig (1992), Banks & Warburton (1991) and Lonergan et al. (1994) describe the structure of the eastern transverse, while Kirker & Platt (1998) and CrespoBlanc & Campos (2001) have worked on the structure of the western Betic Cordillera. Most of these authors indicate that the External Zones rocks are deformed in a thin-skinned fold-and-thrust belt with a detachment level located within the Triassic red bed succession. In the central and eastern Betic Cordillera, the foldand-thrust belt has two deformational fronts, one facing towards the foreland (Iberian Massif) and capped by Upper Tortonian sediments, and the other facing towards the hinterland (Internal Zones) and capped by Upper Burdigalian sedimentary rocks (Garcia-Hernandez et al 1980; de Smet 1984; Banks & Warburton 1991). In

252

E. FERNANDEZ-FERNANDEZ ET AL

the western Betic Cordillera, the fold-andthrust belt has only one deformational front facing towards the foreland (Kirker & Platt 1998; Crespo-Blanc & Campos 2001). Thrusts with a top-to-the-WNW sense of movement deform the Prebetic Zone (Frizon de Lamotte et al 1991; Nebbab 2001). The Intermediate Unit rocks overthrusted by the Subbetic Zone are thrusted, in turn, over the Prebetic rocks (Fig. 2). In the Subbetic Zone of the central Betic Cordillera, folds with a main ENE-WSW trend (which can reach north-south values) are observed, as well as thrust surfaces cutting the folds. The sense of movement of these thrust faults was usually deduced from palaeogeographical studies and from the fold trends (Garcia-Hernandez et al. 1980). This supposed sense of movement was top-to-the-NNW. However, microstructures indicate the existence of several phases of movement younger than the Burdigalian rocks. The first observed movement is top-to-the-west or top-tothe-WSW, while other younger displacements have a top-to-the-NNW sense of movement, and the associated reverse fault surfaces cut the Lower Tortonian rocks (Galindo-Zaldfvar et al. 2000). In the western sector of the Betic Cordillera, the general structure is a NW-vergent fold-and-thrust belt (Kirker & Platt 1998; Crespo-Blanc & Campos 2001) that deforms the sediments of the Early Burdigalian. The folds have a predominantly NE-SW trend, usually verge towards the NW, and have associated axial plane cleavage in the marly rocks (Crespo-Blanc & Campos 2001). These folds are cut by a thrust system with a sole thrust in the Triassic rocks. Kinematic criteria indicate a top-to-the-NW sense of movement for this system (Platt et al 1995; Kirker & Platt 1998; Crespo-Blanc & Campos 2001). The Internal Zones are in thrust contact with the External Zones along the IEZB, which was defined by Lonergan et al. (1994) as a major dextral oblique convergence zone. However, throughout the cordillera, the IEZB corresponds to different fault surfaces, each with its own geometry and kinematics. The IEZB has a main ENE-WSW trend in the eastern and central Betic Cordillera (Fig. 1) and corresponds with a fault zone where the External Zones are emplaced over the Flysch Trough Units and over the Internal Zone rocks. The thrust surface has a NE-SW strike, a NW dip and a top-to-the-SE sense of movement (Paquet 1969; Lonergan 1993; Lonergan et al. 1994). Below the thrust surface of the study area lies the detritic Solana Formation, attributed to the Aquitanian

CONSTRUCTIONAL DEFORMATION IN THE IEZB

and possibly representing a thinned equivalent of the Flysch Trough Units of the western Betics. In the western sector of the cordillera, the IEZB has an arcuate trend in plan view, passing from ENE-WSW trends to northsouth ones (Fig. 1). It is a fault zone along which the Internal Zones are thrust over the External Zones with a top-to-the-NW sense of movement (Balanya & Garcia-Duenas 1987; Platt et al 1995; Kirker & Platt 1998). From structural and palaeomagnetic data, dextral oblique convergence along the IEZB is also inferred in the western Betics (Platt et al 1995). Within the Subbetic Zone, important rotations have been documented around vertical axes with variable amounts and senses of rotation, though most are clockwise (Osete et al. 1988, 1989; Villalain et al. 1992; Platzman 1992, 1994; Allerton et al. 1993; Platt et al. 1995). These rotations seem to be associated with shortening and produced curved trends of the structures in plan view (Allerton 1994), for example, the rotation of the folds and thrust in the region of Sierra Gorda (Platzman 1994). In the study area, the rocks of the Subbetic Zone are folded and the fold trends have a curved shape. This curved pattern is a result of the Neogene shortening, as suggested by the palaeomagnetic study of Allerton et al. (1993), which indicates that most of the study area underwent vertical-axis rotations.

Rock successions The Subbetic rock succession in the study area begins with Early Jurassic to Middle Jurassic limestones and dolostones, and continues with essentially marly rocks that are Middle Jurassic to Aquitanian in age (Fig. 3). We adopt the stratigraphical nomenclature of Rey (1993), who reviewed the previous stratigraphic studies in the area. The lower rocks in the area are 640 m of Hettangian to Lower Pliensbachian limestones and dolostones of the Gavilan Formation (Van Veen 1966; Rey 1993). This formation is overlain by 500 m of white oolitic limestones from the Camarena Formation (Molina 1987), with a Middle Jurassic age. The Gavilan and Camarena Formations define a competent level more than 1000 m thick at the base of the succession. The next formation, the Ammonitico Rosso Formation (Molina 1987), consists of 5 m of red nodular limestones with typical pelagic facies that are Middle-Late Jurassic in age. The Radiolaritas del Charco Formation (Rey 1993) is the lateral facies equivalent of the preceding formation. It is formed by radiolarites

253

and is extremely variable in thickness, reaching up to 100 m. The Lower Cretaceous white marls and marly limestones of the Carretero Formation (Vera et al. 1982) define a discontinuous level over the previous rocks. The next formation is the Lower Cretaceous Fardes Formation (Comas 1978), which comprises dark-green marls, marly clays and clays. The thickness of this formation varies considerably from one zone to another in the study area. It is absent in some places while it can reach more than 1000 m in others. The expansive clays of the Fardes Formation act as a decollement level in the succession. Upper Cretacous to Lower Miocene marly limestone of the Capas Blancas and Capas Rojas Formations (Vera et al. 1982; Martin Algarra 1987) form a more competent level over the clays of the Fardes Formation. Upwards, the Lower Eocene to Aquitanian Barahona Formation (Wittink 1975) is composed of 500 m of green marls and marly limestones. The Capas Blancas, Capas Rojas and Barahona Formations pass laterally into the Conglomerados Calcareos del Puerto Formation (Rey 1993), composed of conglomerates, marls and marly limestones, with abundant olistostromes and slumps. The Middle-Upper Jurassic to Aquitanian formations are usually discontinuous and have lens and wedge shapes with very important variations in thickness. Rey (1998) and FernandezFernandez et al. (2003) have shown that thickness variations were related to a Middle Jurassic to Early Cretaceous rifting stage. The rifting produced highs and basins that are filled by the aforementioned formations, producing onlaps and unconformities. The main highs correspond to the present-day outcrops of the Gavilan and Camarena Formations, while the basins essentially correspond to the outcrops of the marly formations (Figs 4 & 5). All the formations described above are affected by the contractional structures of the area. However, there are several Neogene formations that unconformably overlay the older formations and record only part of the deformation. The lowest of these upper unconformable formations is the Middle Burdigalian Marin Formation (Rey et al. 1990), made up of a calcareous breccia with pebbles from lower formations. The Maiz and the Tala Formations occur in a small outcrop north of Velez Blanco (Fig. 4). The Maiz Formation consists of boulder- and pebble-bearing calcarenites and conglomerates with a marly matrix and a Late Burdigalian Early Langhian age (Aguado & Rey 1996). Overlying the previous rocks is the Tala Formation, composed mainly of light-coloured marls and

254

E. FERNANDEZ-FERNANDEZ ET AL

Fig. 3. Stratigraphic column for the study area.

Fig. 4. (a) Simplified geological map of the study area, with data from Baena et al. (1977) and our own data. Black arrows show the magnetic palaeodeclination determined in the study area by Allerton et al. (1993) and Platt et al. (2003). (Continued}

Fig. 4.

(Continued) (b) Simplified tectonic map of the study area.

Fig. 5. Cross-sections of the study area. See location and legend in Fig. 4. (Continued)

Fig. 5. (Continued)

CONSTRICTIONAL DEFORMATION IN THE IEZB

calcarenites. Geel & Roep (1998) termed it the 'upper sequence of Middle Miocene deposits of Canada del Maiz' and gave it a Late Langhian age. In the study area there are Flysch Trough Unit rocks represented by the Solana Formation (Geel 1973) composed of brown or dark-green marly clays with interlayered sandstones. Geel & Roep (1998) dated this formation as Aquitanian. Structures In the study area, the Subbetic rocks form part of a thrust sheet (the Maimon-La Muela Unit), thrust southward over the Solana Formation (Flysch Trough Units). The main basal thrust surface of this Subbetic unit constitutes, together with the Solana Formation, the IEZB (Figs. 4 & 5). The basal thrust of the Subbetic unit is folded in the eastern part of the study area by a NE-SW-trending antiform, and in the western part of the study area (south of Maria and east of Velez Blanco) by a west-east-trending antiform that generated two tectonic windows (Figs 4, 5a & 6). In the south of the area, the basal thrust superposes the Upper CretaceousLower Eocene Capas Rojas Formation over the Aquitanian Solana Formation. In the Sierra del Maimon (Figs 4, 5a & 6) a small divergent splay of the main basal thrust superposes the Lower Jurassic Gavilan Formation over the Upper Cretaceous-Lower Eocene Subbetic rocks. In this thrust sheet there are five major highs related with the Jurassic-Lower Cretaceous rifting: the Gabar, the Serreta de Guadalupe, the Sierra Larga, the Sierra del Pericay and the Sierra del Gigante (Fig. 4). There are also five major basins: the Northern Basin, the Rambla Seca Basin, the Eastern Gabar Basin, the Arroyo de Taibena Basin and the Zarzilla de Ramos Basin (Fig. 4). Most of these basins are associated with synsedimentary normal faults and folds that controlled the geometry of the deposits (Rey 1998; Fernandez-Fernandez et al 2003). The Rambla Seca Basin corresponds to a half-graben (Figs 5 & 7), while the Arroyo Taibena Basin is a graben (Fig. 5). The Eastern Gabar Basin, however, is limited in its northern border by an antiform and the basin is a synform (central part of cross-section 2-2', Fig. 5b). The geometry of the formations in the area is not tabular and most of the formations are discontinuous. The normal faults associated with these basins are capped by sediments of different ages; for example, the upper part of the Fardes Formation caps the fault associated with the Rambla Seca Basin, which is AptianAlbian in age (Fernandez-Fernandez et al.

259

2003). In other places, as in the Arroyo Taibena Basin, the faults can be active up to the Palaeogene. All these basins and highs are related to a rifting stage developed during Late Jurassic-Early Cretaceous times (Rey 1998; Nieto et al 2001; Vilas et al 2001; FernandezFernandez et al 2003). Several small faults associated with this extensional deformation have been measured and their orientations are shown in Figure 8a; one characteristic of the normal faults is that they usually have a very high dip. Folds and thrusts deform this succession with the highs and basins. The folds have south and east vergences and affect the Lower Jurassic to Aquitanian rocks, and also part of the Lower Miocene Marin Formation. The fold axial traces change in trend from NNE-SSW in the northeastern part of the study area to WSWENE in the rest of the area (Figs 4 & 7). The folds usually vary in geometry from closed to open, and are mainly non-cylindrical and doubly plunging (Fig. 9). The geometry of the folds varies through the Maimon-La Muela Unit. In the northern part of the area, coinciding with the Northern Basin where the Fardes Formation has the greatest thickness, the folds are closed with overturned limbs. In this area, the folds verge towards the south, SE and east (Fig. 5a, section 1-10 and have amplitudes that vary from 500 to 700 m in the major folds and between 50 and 200 m in the intermediate folds. The wavelengths also vary from 1000 to 1500m, and from 200 to 600m, respectively (Fig. 5b, section 2-2'). In the central part of the area, where the Fardes Formation is usually thin or absent, the folds are open or very open, and they do not show overturned limbs. In this area, the major folds have amplitudes around 150m, while the wavelengths vary from 1000 to 700 m. The minor folds have amplitudes around 50 m, while the wavelengths oscillate between 200 and 350 m. The axial surface is subvertical or dips steeply towards the NW. In the southernmost part of the area, the folds also have overturned limbs. In the upper thrust mass located between Maria and Velez Blanco (Fig. 6a), the two southern folds that developed in the Gavilan and Camarena Formations show subvertical or reverse limbs and axial surfaces dipping towards the north, while the northern fold has a subvertical axial surface. In the Northern Basin, the geometry of the folds seems to be independent of the geometry of the formations. In the rest of the area, however, the major folds seem to be related to the thickness of the succession. The major antiforms develop mainly in the highs of the Gavilan and

Fig. 6. (a) Detailed geological map of the region around Maria and Velez Blanco showing the small tectonic windows of the Solana Formation and the capping of the Subbetic basal thrust by the Middle Burdigalian Maiz Formation. (Continued)

Fig. 6. (Continued) (b, c) Cross-sections of the tectonic window and of the frontal splay of the Subbetic thrust system.

Fig. 7. (a) Detailed map of the Rambla Seca Basin. (Continued)

CONSTRICTIONAL DEFORMATION IN THE IEZB

263

Camarena Formations, while the major synforms occur in the basins where the Upper Jurassic to Palaeogene rocks are extremely thick. Associated with these folds, an axial plane foliation is developed only in the Capas Blancas and Capas Rojas Formations (Fig. lOb). It has a disjunctive cleavage, with surfaces separated essentially by 1 or 2 cm wide microlithons of the marly limestones. On several of these surfaces, stylolites can be observed, but in other surfaces, the cleavage planes are decorated by a filling of calcite crystals. In the outcrop of the 'Cueva del Ambrosio' (Fig. 11 a), a progressive unconformity of the Marin Formation (Middle Burdigalian) over the Upper Cretaceous to Palaeogene Capas Rojas Formation can be observed. In this unconformity it can be seen how, in the lowest beds of the Marin Formation, the folds are more open than in the Capas Rojas Formation and tighter than in the upper beds of the Marin Formation (Fig. llc-f, cross-sections A-A7 and B-B'). Moreover, in the same structure, the plunge of the fold decreases in the higher beds of the Marin Formation (Fig. llc-g, cross-sections A-A r and B-B' and stereonet diagram). These features indicate that, although the folding began before the deposition of the Marin Formation, the tightening of the folds continued during the deposition of the Marin Formation, as well as the increase in the plunge of the hinges and the transition from cylindrical folds to non-cylindrical doubly plunging folds. The synsedimentary normal faults and the folds are cut by a thrust system that superposes the Maimon-La Muela Unit over the Flysch Trough Units. This thrust system constitutes the IEZB in this area. The trace of the thrust surfaces has a mean ENE-WSW trend in the south of the area. However, the trace of the fault system has a strong inflection in the eastern part of the study area that corresponds to a NNE-SSW antiform that does not affect the rocks of the base of the Solana Formation. The basal thrust also crops out in two tectonic windows east and west of Velez Blanco (Figs 4, 5 & 6) in a very open east-west antiform. The basal thrust surface has an associated fault rock that is 50-70 cm thick. The striations in the fault rock have trends from N20°E to N150°E (Fig. 8). The superposition of the rocks and most of the kinematic criteria (mainly S-C structures in the fault gouge) show a top-to-the-SSE sense of movement (Fig. 10). These results are very similar to those obtained by Lonergan et al (1994). However, we can distinguish two areas with slightly different trends and senses of movement along the thrust fault, one in the

264

E. FERNANDEZ-FERNANDEZ ET AL

Fig. 8. Diagrams of the orientation of the striations and minor fault planes in the fault rocks of: (a) a synsedimentary Mesozoic normal fault, (b) the main thrust surface, (c) the main thrust surface in the western area, and (d) the main thrust surface in the eastern area. Wulffnet, lower hemisphere. Stars, striae of normal faults; triangles, striae of the thrust faults; squares, striae of the dextral strike-slip faults; diamonds, striae of the sinistral strike-slip faults; circles, striae with unknown sense of movement. The rose diagrams indicate the main direction of transport in the thrust surface.

Fig. 9. Map of the area with diagrams of the orientation of poles of bedding in different sectors. Schmidt net, lower hemisphere.

266

E. FERNANDEZ-FERNANDEZ ET AL

Fig. 10. (a) Minor-scale fold in the Capas Rojas Formation, (b) Axial-plane foliation in the Capas Rojas Formation. (Continued)

west (close to Velez Rubio) and the other in the eastern part of the study area. In the western region, the main trend of the striations varies from N145°E to N200°E (Fig. 8c). In this western region, the striations with trends from N145°E to N170°E mainly

have a top-to-the-SSE sense of movement of the hanging wall. However, when the striations have a north-south to N200°E trend, the associated sense of movement of the hanging wall is towards the north (Fig. 8c). In the eastern region, the main trends of the striations are

CONSTRICTIONAL DEFORMATION IN THE IEZB

267

Fig. 10. (Continued) (c) Detail of the rock fault in the main thrust surface; south is on the left, (d) Detail of the surface and the fault rock in the klippen south of Maria.

east-west to N190°E, the associated sense of movement of the hanging wall is towards the ESE and south, and there are no senses of movement towards the north (Fig. 8d). The upper thrust located in the southwestern part of the study area between the villages of Maria and Velez Blanco (Fig. 6) superposes the Gavilan Formation over the Capas Rojas and Barahona Formations essentially with a northsouth movement trend. The striations measured on the fault surface range between N160°E and N190°E (Fig. 8c) and the S-C structures observed in the associated fault zone generally indicate a sense of movement of the hanging wall towards the north in contrast with the whole movement of the thrust system towards the SE (Fig. 10). The basal thrust cuts the Marfn Formation in the eastern region, north of the 'embalse de

Valdeinfierno' (Valdeinfierno reservoir) (Fig. 7), and in the Zarcilla de Ramos Basin (Fig. 12), indicating that the thrust activity is later than the Middle Burdigalian. In the western tectonic window, the calcarenites and conglomerates of the Maiz Formation, Late Burdigalian-Early Langhian in age (Aguado & Rey 1996), unconformably cover the basal thrust and directly overlie the marly clays and sandstones of the Solana Formation (Fig. 6). In the hanging wall of this basal thrust, there are several reverse faults with similar kinematics, which we interpret as associated with the main thrusting episode. The reverse faults have different characteristics depending on their position around the curved fold trace. The reverse faults located in the area where the folds have ENEWSW trends usually have ENE-WSW strikes

Fig. 11. (a) Aerial photograph of the 'Cuevas del Ambrosio' area, (b) A sketch of the cartography of the area.

CONSTRUCTIONAL DEFORMATION IN THE IEZB

269

Fig. 11. (Continued) (c-f) Images and interpretative cross-sections A-A' and B-B' on the map, which illustrate the geometry of the internal unconformity in the Marin Formation, (g) Diagram of the orientation of bedding and hinges of folds in the outcrop. Wulff net, lower hemisphere. •, poles to bedding in the Capas Rojas Fm (CCR); •, hinge of syncline in the Capas Rojas Fm (CCR); A, hinge of syncline in the lower beds of Marin Fm (TM); *, hinge of syncline in the higher beds of Marin Fm (TM).

270

E. FERNANDEZ-FERNANDEZ ET AL

Fig. 12. (a) Geological map of the breached duplex in the eastern region of the study area, (b) Cross-section of the structure.

CONSTRICTIONAL DEFORMATION IN THE IEZB

and 2 or 4 km of lateral development. They normally dip towards the north and cut the reverse limbs of the folds. They show small slips and striations with N150°E trends. The kinematic criteria generally indicate a top-to-the-SE sense of movement. One of these faults is located SW of Serreta de Guadalupe (Fig. 7a, b, thrust in cross-section A-A'). This reverse fault is folded by an open synform and cuts the succession of a half-graben basin filled with the Fardes Formation and the Barahona Formation. The thrust fault is located in the base of the Barahona Formation and has striations with a N140°E trend and a top-to-the-SE sense of movement. This fault does not seem to cut the normal fault that defined the half-graben. The reverse faults developed in the NE part of the study area where the folds have a NNESSW trend and constitute a duplex that we term here the Valdeinfierno Duplex (Fig. 7a, d, crosssection C-C). Allerton et al (1994) have studied this duplex north of the study area in the Barranco Salada section. The roof thrust of the duplex crops out in five klippes (Fig. 7) and is cut by a NNE-SSW normal fault zone. The roof thrust superposes the oolitic limestones of the Camarena Formation over the Middle-Upper Jurassic to Aquitanian Formations. There are two horses; one of them seems to affect the whole succession, while the smaller horse affects only the Capas Rojas and the Marin Formation. The fact that the Marin Formation is cut and deformed in this duplex indicates that the reverse faults are later than the Middle Burdigalian, and probably have the same age as the basal thrust system. In the eastern part of the study area, south of Zarzilla de Ramos, the basal thrust surface of the Maimon-La Muela Unit is cut out of sequence by several slices of a lower thrust system. This process has originated breaching horses (McClay 1992), which produced a breaching duplex (Butler 1987) (Fig. 12). This latter thrust system repeats the basal thrust surface of the Maimon-La Muela Unit, leading to repetition of the Capas Rojas and Barahona Formations (hanging wall) and the Solana Formation (footwall). Within the Solana Formation, there are two folds with a N50°E trend and reverse folds that can be interpreted as having been generated by two blind thrusts. The trend of these folds suggests that this second thrust system had a SE sense of movement for the hanging wall. Other structures in the study area are NWvergent folds in the northwestern and southern regions. These folds have N80°E subhorizontal hinges and are clearly asymmetric, with subvertical northern limbs and very long subhorizontal

271

southern limbs. They deform the thrust surfaces and previous folds and also affect the whole Marin Formation in the northwestern region of the study area (cross-section 1-1', Fig. 5a). Discussion Regional constraints In the Maimon-La Muela Unit, the style of the Neogene folding depends on the succession of rocks affected. In the northwestern part of the study area, where the Fardes Formation is well developed, this formation seems to act as a decollement level and the folds usually have reverse limbs and greater amplitudes than in the rest of the area. In the central part of the study area, where the Fardes Formation is only locally present, the folds are more open and there are no reverse limbs. However, in the southern part of the Maimon-La Muela Unit, the folds also have reverse limbs. With the exception of these southern folds, the location of the folds seems to be controlled by the Mesozoic and Palaeogene palaoegeography, and the major antiforms are located in the ancient highs of limestones and dolostones of the Gavilan and Camarena Formations, while the major synforms are located in the ancient basins. When the highs correspond with a synsedimentary antiform and the basins with a synsedimentary synform, the Neogene folds produced a tightening of these folds. These features can be explained by the fact that the succession in the highs is very reduced, with frequent omissions and onlaps, while in the basins the succession is very thick and the omissions are less frequent. Thus, the basins and highs defined very open folds that were tightened during the Neogene shortening. The folding and shortening in the basins was accommodated by different mechanisms depending on the previous shape of the basin. Where the basin is a half-graben, like the Rambla Seca Basin, the footwall (Serreta de Guadalupe) became an open antiform, and the ancient normal fault probably rotated to become steepened. In the hanging wall, the wedge-shaped fill of the half-graben basin became folded by a synform tighter than the fold in the footwall. Owing to the asymmetry of this fill, the resulting fold is asymmetric, with one well-developed limb and another limb that is nearly absent (Fig. 7). The space problems in the nucleus of the synform may be accommodated by an out-of-syncline thrust - or 'a thrust fault that nucleates and propagates out from the core of a syncline... and may not necessarily be linked to other thrust' (McClay 1992) - such as the

272

E. FERNANDEZ-FERNANDEZ ET AL

fault observed SW of the Serreta de Guadalupe (Fig. 7a, b, cross-section A-A'). The fault has a direction of movement (N140°E) that is oblique to the fold hinge, suggesting that folding was generated by a mechanism of flexural folding, but with a transpressive component that accounts for the obliquity of the striations. In the graben of the Arroyo de Taibena Basin, the conjugated normal faults seem to acquire a steeper dip and the filling of the basin became folded with sub vertical axial planes. The Neogene folding affects the top of the Barahona Formation, dated as Aquitanian, while the Middle Burdigalian Marin Formation developed the progressive unconformities associated with the folds. These features suggest that the folds began in the Early Burdigalian and were tightened during the Middle Burdigalian; and that the folding concluded in the Late Burdigalian, as the folds are cut by the thrust system, which is capped by the Maiz Formation. The tightening of the folds was associated with the loss of cylindricity (in the sense that a folded surface can be described by the parallel movement of a line in space) because the fold hinges became curved within the axial plane and the fold became a doubly plunging one. The evolution from straight to curved hinges suggests the existence of shortening parallel to the hinge line. This feature associated with the shortening normal to the hinges, indicated by the tightening of the folds, allows us to propose that the horizontal plane experienced area reduction strain during the Middle Burdigalian. Pattern of the vertical-axis rotations Allerton et al. (1993) have determined several palaeomagnetic declinations in the eastern External Zones of the Betic Cordillera. Among these, there are six declinations located within the study area (Fig. 4). Three of these palaeodeclinations are located in the Sierra del Pericay, in a region where the main structures have a NNE-SSW trend. They show counterclockwise rotations with values of -12° ± 13.8°, -8° ± 7.4° and -8° ± 12.9°. There are two other declinations obtained in the northern border of Sierra Larga (Fig. 4), where the structures have an ENEWSW trend. They show clockwise rotations with values of 64° ± 9.8° and 80° ± 5.8°. Allerton et al (1993) have also determined a declination in the upper thrust surface south of Maria (Fig. 4a) that indicates a clockwise rotation of 137° ± 9°. In addition, Platt et al (2003) have determined several palaeomagnetic declinations in the region, one of which, from the Sierra del Pericay, shows a counterclockwise rotation of

-15° ± 5° (Fig. 4). Another location, in the Sierra Larga, shows a clockwise rotation of 62° ± 7.3°, and yet another in the Gabar indicates an average clockwise rotation of 63° ± 5.5°. Platt et al. (2003) have also determined a palaeomagnetic declination in the upper thrust surface south of Maria (Fig. 4a) with a clockwise rotation of 74° ±12°. These palaeomagnetic declinations have all been determined in the Ammonitico Rosso Formation, and the age of the rotations is assumed to be Neogene (Allerton etal 1993; Allerton 1994). All of these data indicate that the fold axial trace curvature is the result of later rotations with a vertical axis that deform the fold trends. Where the structures have a NNE-WSW trend, the declination data indicate a counterclockwise rotation between —8° and —15°. These data suggest that the original orientation of the folds was approximately north-south and the asymmetric folds were east-verging. In the areas where the folds have present-day ENE-WSW trends, the declination data indicate clockwise rotations between 62° and 80°, which also suggest that the original orientations of the folds were between north-south and NNE-SSW. Therefore, the folds seem to have been generated with an approximately northsouth trend and were later folded. The thrust system that superposed the Subbetic Unit over the Flysch Trough Units affects the Middle Burdigalian Marm Formation, which is cut by several thrust surfaces. Moreover, the Upper Burdigalian-Lower Langhian Maiz Formation capped the basal thrust surface to the north of Velez Blanco (Fig. 6). These features indicate that the age of the thrust system is Middle to Late Burdigalian. With respect to the transport direction of this thrust system, we can observe top-to-the-north senses of movement in the thrust surfaces of the western part of the study area. However, in both the western and the eastern areas of the basal thrust system, we also note a top-to-the-SE sense of movement. This top-to-the-SE sense of movement agrees with the observations of Lonergan et al (1994) regarding the kinematics of the Subbetic basal thrust NW of Sierra Espuna, which is of the same age, and with the kinematics of the IEZB in Velez Rubio (Lonergan et al 1994). It may be that the thrust system has a mean top-to-the-SE sense of movement and the top-to-the-north sense of movement is a later deformation that developed only in the western part of the study area. The top-to-the-SE sense of movement shows minor differences (20° to 30°) in the transport direction from the western to the eastern outcrops in the basal thrust (Fig. 9), which can be explained if part of the vertical-axis rotations that

CONSTRICTIONAL DEFORMATION IN THE IEZB

affected the folds were accommodated along the thrust plane during the movement of the thrust mass. This vertical-axis rotation seems to be younger than the end of folding, Middle Burdigalian in age. Rotation may be partially coeval with the movement of the Subbetic thrust system, as the striations in the main basal surface do not show the same rotation pattern as the folds. The arcuate pattern suggests that the approximate main shortening direction during the vertical-axis rotation stage was approximately NE-SW, while the main extensional direction was approximately NW-SE. However, the slip directions of the basal thrust surface suggest a main shortening direction that seems to be NW-SE, and that may indicate a clockwise rotation of the horizontal strain axis during the Middle and Late Burdigalian. The presence of a breached duplex in the eastern part of the study area indicates the presence of a lower thrust system below the Subbetic thrust system. This lower thrust system is obviously younger, as it cut the IEZB. This age relationship indicates that the thrust systems have a piggyback sequence. The folds developed in the footwall rocks and the cut-off lines show a NW-SE sense of movement for the hanging wall of this thrust system. The top-to-the-north senses of movement recorded in the various thrust surfaces in the western part of the study area may be a consequence of the late north-vergent asymmetric folds that could have produced secondary movements towards the north in the ancient fractures. These folds record late NNW-SSE shortening. Evolution of the structures and implication for coupling and decoupling The evolution of the contractional structures can be summarized as follows. In a first stage during the Early Burdigalian, the Subbetic rocks in the study area underwent a folding stage with eastwest shortening that produced nearly cylindrical folds. The geometry and location of these folds were controlled by a previous extensional stage (Fernandez-Fernandez et al 2003). The second stage corresponds to the tightening of the folds during the Middle Burdigalian associated with the loss of cylindricity. The vertical-axis rotation of the structures followed this tightening stage; and both deformations indicate conditions with reduction in area. The orientation of the shortening direction during the vertical-axis rotation stage was approximately NE-SW. The following stage, during the transition from Middle to Late

273

Burdigalian times, involved the development of the thrust system that superposed the Subbetic Zone over the Flysch Trough Units. The N80°E north-vergent asymmetric folds that record NNWSSE shortening represent the last stage (Fig. 13). The evolution of the structures of the Subbetics described here can be explained by an initial episode of folding with north-south trends during the Early Burdigalian, which may correspond with the convergence between the External and the Internal Zones, with east-west shortening. The second stage corresponds to transpressive dextral deformation during the Middle to Late Burdigalian that began with the tightening and posterior rotation of the previous folds, followed by the activity of the thrust systems. This transpression is evidenced by the rotation of the main axis of the horizontal strain ellipse as well as the area reduction during the vertical-axis rotations. The cause of this transition from east-west pure convergence to dextral transpression may be the rotation of the front of the Internal Zones, probably due to the extensional deformation active in the Alboran Sea during this period (Galindo-Zaldivar et al. 1989; Gonzalez-Lodeiro et al. 1996; Lonergan & White 1997). This dextral transpression corresponds to the motion of the IEZB during the Burdigalian. The most important implication of this case study for the deformation at convergent margins is that, even in the uppermost level of the crust, the partitioning of the normal shortening and the strike-slip component of the transpression is not complete, and part of the strike-slip deformation is accommodated by internal deformation of the rocks in the fold-and-thrust belt and even by slip in the thrust surfaces. This internal deformation is mostly accommodated by heterogeneous vertical-axis rotation. This case study favours the presence of a partial coupling in the upper levels of the lithosphere. Conclusions The Neogene compression regime affecting the Subbetic Zone in the eastern Betic Cordillera produced folds during the Burdigalian whose geometry was controlled by the existence of several highs and basins produced during a previous extensional stage. The shortening of these basins caused the reorientation of the ancient normal faults and the folding of the basins. The folds have associated progressive unconformities of Middle Burdigalian age in the Marin Formation, which recorded the tightening of the folds and their loss of cylindricity. The folds have a present-day curved pattern in plan view

274

E. FERNANDEZ-FERNANDEZ ET AL.

Fig. 13. Sketch of the evolution of the study area during the Burdigalian. (a) East-west shortening during the Early Burdigalian. (b) Loss of the cylindricity and tightening of the folds developed in the previous stage; vertical-axis rotation of the folds and beginning of the thrusting during dextral transpressive deformation with area reduction during the Middle Burdigalian. (c) Development of the breaching duplex during the same dextral transpression in the Late Burdigalian.

that is the result of vertical-axis rotations. This curved pattern is evidence of a period of NESW shortening, probably with a reduction in area. The folds are located in the hanging wall

of a thrust system with a top-to-the-SE sense of movement that was active during the Middle to Late Burdigalian. This thrust system represents the IEZB, since it superposes the Subbetic

CONSTRICTIONAL DEFORMATION IN THE IEZB

rocks over the Flysch Trough Units. The thrust surface is cut by a lower thrust system with the same kinematics, producing a breached duplex. The north-vergent asymmetric folds could account for the reworking of the ancient thrust surfaces with new top-to-the-north senses of movement. This evolution can be explained by a first stage of east-west pure convergence between the External and the Internal Zones during the Early Burdigalian, followed by dextral transpression that produced the rotation of the previous structures in the hanging wall, with a reduction in area. In summary, this case study shows how the vertical-axis rotations associated with oblique convergence are accommodated in the uppermost levels of the crust by the deformation with area reduction of previous structures (folds), and by thrusting. We would like to thank Jean Louise Sanders for the English version of this chapter and Jesus Rodero for help with the drawings. This work was supported by the 'Grupo de Investigation de la Junta de Andalucia: Geologia Estructural y Tectonica' and by the CICYT project BET2000-1490-C02-01.

References AGUADO, R. & REY, J. 1996. Consideraciones sobre la edad del techo de las calizas ooliticas del Jurasico medio del Subbetico Interne oriental (Cordillera Betica). Geogaceta, 20, 35-38. ALLERTON, S. 1994. Vertical axis rotation associated with folding and thrusting: an example from the eastern Subbetic of southern Spain. Geology, 22, 1039-1042. ALLERTON, S., LONERGAN, L., PLATT, J.P., PLATZMAN, E.S. & MCCLELLAND, E. 1993. Palaeomagnetic rotations in the eastern Betic Cordillera, Southern Spain. Earth and Planetary Science Letters, 119, 225-241. ALLERTON, S., REICHERTER, K. & PLATT, J.P. 1994. A structural and palaeomagnetic study of a section through the eastern Subbetic, southern Spain. Journal of the Geological Society, London, 151, 659-668. ALONSO-ZARZA, A.M., ARMENTEROS, I. et al. 2002. Tertiary. In: GIBBONS, W. & MORRENO, T. (eds) The Geology of Spain. Geological Society, London, 293-334. AZEMA, J., FOUCAULT, A. et al. 1979. Las Microfacies del Jurasico y Cretdcico de las Zonas Externas de las Cordilleras Beticas. Secretariado de Publicaciones Universidad de Granada, Granada. BAENA, J., TORRES, J., GEEL, T. & ROEP, T.B. 1977. Mapa Geologico y Memoria Explicativa, N° 952 (Velez-Blanco). Institute Geologico y Minero de Espana, Madrid. BALANYA, J.C. & GARCIA-DUENAS, V. 1987. Les directions structurales dans le Domaine d'Alboran

275

de part et d'autre du Detroit de Gibraltar. Comptes Rendus de I'Academie des Sciences, Paris, 304, 929-932. BANKS, C.J. & WARBURTON, J. 1991. Mid-crustal dettachment in the Betic system of southeast Spain. Tectonophysics, 191, 275-289. BECK, M.E. 1983. On the mechanism of tectonic transport in zones of oblique subduction. Tectonophysics, 93, 1-11. BLUMENTHAL, M. 1927. Versuch einer tektonischen Gliederung der betischen Cordilleren von Central und Siidwest Andalusien. Eclogae Geologae Helvetiae, XX, 487-592. BUTLER, R.W.H. 1987. Thrust successions. Journal of the Geological Society, London, 144, 619-634. CALVO, J.M., CUEVAS, J. & TUBIA, J.M. 2001. Preliminary palaeomagnetic results on Oligoceneearly Miocene mafic dykes from southern Spain. Tectonophysics, 332, 333-345. CHALOUAN, A. & MICHARD, A. 1990. The Ghomarides nappes, Rif coastal range, Morocco: a variscan chip in the Alpine belt. Tectonics, 9, 1565-1583. COMAS, M.C. 1978. Sobre la geologia de los Monies Orientales: sedimentacion y evolucion paleogeogrdfica desde el Jurasico al Mioceno inferior (Zona Subbetica, Andalucia). Doctoral thesis, Universidad de Bilbao, Spain. COMAS, M.C., PLATT, J.P., SOTO, J.I. & WATTS, A.B. 1999. The origin and tectonic history of the Alboran Basin: insights from leg 161 results. In: ZAHN, R., COMAS, M.C. & KLAUS, A. (ed.) Proc. ODP, Scientific Results, 161, 555-579. Ocean Drilling Program, College Station, TX. CRESPO-BLANC, A. & CAMPOS, J. 2001. Structure and kinematics of the south Iberian paleomargin and its relationship with the Flysch Trough units: extensional tectonics within the Gibraltar Arc fold-andthrust belt (western Betics). Journal of Structural Geology, 23, 1615-1630. DE RUIG, M.J. 1992. Tectono-sedimentary evolution of the Prebetic fold belt of Alicante. Doctoral thesis, Free University of Amsterdam. DE SMET, M.E.M. 1984. Wrenching in the External Zone of the Betic Cordilleras, southern Spain. Tectonophysics, 107, 57—79. FALLOT, P. 1945. Estudios Geologicos en la Zona Subbetica entre Alicante y el Rio Guadiana Menor. Memoria del Instituto Lucas Mallada, CSIC, Madrid. FALLOT, P. 1948. Les Cordilleres betiques. Estudios Geologicos, IV, 259-279. FERNANDEZ-FERNANDEZ, E., JABALOY, A. & GONZALEZ-LODEIRO, F. 2003. Middle Jurassic to Cretaceous extensional tectonics and sedimentation in the Eastern External Zone of the Betic Cordillera. In: MCCANN, T. (ed.) Tracing Tectonic Deformations Using the Sedimentary Record. Geological Society, London, Special Publications. 208, 29-53. FITCH, T.J. 1972. Plate convergence, transcurrent faults, and internal deformation adjacent to Southeast Asian and Western Pacific. Journal of Geophysical Research, 77, 4432-4461.

276

E. FERNANDEZ-FERNANDEZ ET AL

FOUCAULT, A. 1960. Decouverte d'une nouvelle unite tectonique sous le massif subbetique de la Sierra Sagra (Andalousie). Comptes Rendus de I'Academic des Sciences, Paris, 250, 20-38. FOUCAULT, A. 1962. L'Unite du Rio Guardal (Prov. de Grenade, Espagne) et les liaisons entre prebetique et subbetique. Bulletin Societe Geologique de France, 4, 446-452. FRIZON DE LAMOTTE, D., ANDRIEUX, J. & GUEZOU, J.C. 1991. Cinematique des chevauchements Neogenes dans 1'arc Betico-Rifain: discussion sur les modeles geodynamiques. Bulletin Societe Geologique de France, 62, 611—626. GALINDO-ZALDIVAR, J., GONZALEZ-LODEIRO, F. & JABALOY, A. 1989. Progressive extensional shear structures in a detachment contact in the Western Sierra Nevada (Betic Cordilleras, Spain). Geodinamica Acta, 3, 73-85. GALINDO-ZALDIVAR, J., RUANO, P., JABALOY, A. & LOPEZ-CHICANO, M. 2000. Kinematics of faults between Subbetic Units during the Miocene (central sector of the Betic Cordillera). Comptes Rendus de VAcademie des Sciences, Paris, 331, 811-816. GARCIA-DUENAS, V. 1967. La Zona Subbetica al Norte de Granada. Doctoral thesis, Universidad de Granada, Spain. GARCIA-HERNANDEZ, M., LOPEZ GARRIDO, A.C., RIVAS, P., SANZ DE GALDEANO, C. & VERA, J.A. 1980. Mesozoic paleogeographic evolution the External Zones of the Betic Cordillera. Geologie en Mijnbouw, 59, 155-168. GARCIA-HERNANDEZ, M., LOPEZ-GARRIDO, A.C., MARTIN-ALGARRA, A., MOLINA, J.M., RUIZORTIZ, P.A. & VERA, J.A. 1989. Las discontinuidades mayores del Jurasico de las Zonas Extenas de las Cordilleras Beticas: analisis e interpretation de los ciclos sedimentarios. Cuadernos de Geologia Iberica, 13, 35-52. GEEL, T. 1973. The geology of the Betic of Malaga, the Subbetic and the zone between these two units in the Velez Rubio area (Southern Spain). GUA Papers of Geology, I, 5. GEEL, T. & ROEP, T.B. 1998. Oligocene to middle Miocene basin development in the Eastern Betic Cordilleras, SE Spain (Velez Rubio corridor - Espuna): reflections of West Mediterranean plate-tectonic reorganizations. Basin Research, 10, 325-343. GONZALEZ-LODEIRO, F., ALDAYA, F., GALINDOZALDIVAR, J. & JABALOY, A. 1996. Superimposition of extensional detachments during the Neogene in the internal zones of the Betic cordilleras. Geologische Rundschau, 85, 350-362. JONES, S., TURNER, P. & TURNER, J. 1996. Palaeomagnetism and basin analysis of the Serrania del Interior. American Association of Petroleum Geologists, 80, 1304. KIRKER, A.I. & PLATT, J.P. 1998. Unidirectional slip vector in the western Betic Cordilleras: implications for the formation of the Gibraltar arc. Journal of the Geological Society, London, 155, 193-208.

LONERGAN, L. 1993. Timing and kinematics of deformation in the Malaguide Complex, Internal Zone of the Betic Cordillera, southeast Spain. Tectonics, 12, 460-476. LONERGAN, L. & PLATT, J.P. 1995. The MalaguideAlpujarride boundary: a major extensional contact in the Internal Zone of the eastern Betic Cordillera, SE Spain. Journal of Structural Geology, 17, 1655-1671. LONERGAN, L. & WHITE, N. 1997. Origin of the Betic-Rif mountain belt. Tectonics, 16, 504-522. LONERGAN, L., PLATT, J.P. & GALLAGHER, L. 1994. The internal-external zone boundary in the eastern Betic Cordillera, SE Spain. Journal of Structural Geology, 16, 175-188. MARTIN ALGARR A, A. 1987. Evolucion geologica alpina del contacto entre las Zonas Internas y las Zonas Externas de las Cordilleras Beticas (Sector central y occidental). Doctoral thesis, Universidad de Granada, Spain. MCCAFFREY, R. 1992. Oblique plate convergence, slip vectors, and forearc deformation. Journal of Geophysical Research, 97, 8905-8915. McCLAY, K.R. 1992. Glossary of thrust tectonics terms. In: McCLAY, K.R. (ed.) Thrust Tectonics. Chapman & Hall, London, 1017-1048. MCQUARRIE, N. 2002. Initial plate geometry, shortening variations, and evolution of the Bolivian orocline. Geology, 30, 867-870. MOLINA, J.M. 1987. Analisis de fades del Mesozoico en el Subbetico Externo (Provincias de Cordoba y sur de Jaen). Doctoral thesis, Universidad de Granada, Spain. NEB BAD, F. 2001. Le prisme orogenique Prebetique (Sud-est de I'Espagne), evolution cinematique et coupes equilibrees. Doctoral thesis, Universite de Paris IX, France. NIETO, L.M., DE GEA, G.A., AGUADO, R., MOLINA, J.M. & RUIZ-ORTIZ, P.A. 2001. Procesos sedimentarios y tectonicos en el transito Jurasico/ Cretacico: precisiones bioestratigraficas (Unidad del Ventisquero, Zona Subbetica). Revista de la Sociedad Geologica de Espana, 14, 35-46. OSETE, M.L., FREEMAN, R. & VEGAS, R. 1988. Preliminary paleomagnetic results from the Subbetic Zone, Betic Cordillera, southern Spain: kinematic and structural implications. Physics of the Earth and Planetary Interiors, 52, 283-300. OSETE, M.L., FREEMAN, R. & VEGAS, R. 1989. Paleomagnetic evidence for block rotations and distributed deformation of the Iberian-African plate boundary. In: Kissel, C. & Laj, C. (eds.) Paleomagnetic Rotations and Continental Deformation. Kluwer Academic, Dordrecht, 381-391. PAQUET, J. 1969. Etude geologique de 1'ouest de la province de Murcie (Espagne). Memoires de la Societe Geologique de France, 3, 1-270. PERCONIG, E. 1962. Sur la constitution geologique de 1'Andalousie occidentale, en particulier du bassin du Guadalquivir (Espagne meridionale). In: Livre a la memoire du Professeur Paul Fallot. Bulletin Societe Geologique de France, 1, 229-256.

CONSTRICTIONAL DEFORMATION IN THE IEZB PLATT, J.P. 1993. Mechanics of oblique convergence. Journal of Geophysical Research, 98, 1623916256. PLATT, J.P. 2000. Calibrating the bulk rheology of active obliquely convergent thrust belts and forearc wedges from surface profiles and velocity distributions. Tectonics, 18, 529-548. PLATT, J.P., ALLERTON, S., KIRKER, A. & PLATZMAN, E.S. 1995. Origin of the western Subbetic arc (southern Spain): palaeomagnetic and structural evidence. Journal of Structural Geology, 17, 765-775. PLATT, J.P., SOTO, I.I., WHITEHOUSE, M.J., HURFORD, AJ. & KELLEY, S.P. 1998. Thermal evolution, rate of exhumation and tectonic significance of metamorphic rocks from the floor of the Alboran extensional basin, western Mediterranean. Tectonics, 17,671-689. PLATT, J.P., ALLERTON, S., KIRKER, A., MANDEVILLE, C., MAYFIELD, A., PLATZMAN, E.S. & RIMI, A. 2003. The ultimate arc: differential displacement, oroclinal bending and vertical axis rotation in the formation of the External Betic Rif arc. Tectonics, 22, 1017-1048. PLATZMAN, E.S. 1992. Paleomagnetic rotations and the kinematics of the Gibraltar Arc. Geology, 20, 311-314. PLATZMAN, E.S. 1994. East-west thrusting and anomalous magnetic declinations in the Sierra Gorda, Betic Cordillera, southern Spain. Journal of Structural Geology, 16, 11-20. PLATZMAN, E.S., PLATT, J.P., KELLEY, S.P. & ALLERTON, S. 2000. Large clockwise rotations in an extensional allochthon, Alboran Domain (southern Spain). Journal of the Geological Society, London, 157, 1187-1197. REY, J. 1993. Andlisis de la cuenca subbetica durante el Jurdsico y el Cretdcico en la transversal Caravaca Velez Rubio. Doctoral thesis, Universidad de Granada, Spain. REY, J. 1998. Extensional Jurassic tectonism of an eastern Subbetic section (southern Spain). Geological Magazine, 135, 685-697. REY, J., ANDREO, B., GARCIA-HERNANDEZ, M., MARTIN-ALGARRA, A. & VERA, J.A. 1990. The

277

Lower Jurassics Lithiotis facies north of Velez Rubio (Subbetic Zone). Revista de la Sociedad Geologica de Espana, 3, 199-212. SANZ DE GALDEANO, C., MARTIN ALGARRA, A., VERA, J.A. & RIVAS, P. 1993. Reply to comments on 'Structure and palaeogeography of the External Betic Cordillera, southern Spain'. Marine and Petroleum Geology, 10, 518-521. TEYSSIER, C. & TIKOFF, B. 1998. Strike-slip partitioned transpression of the San Andreas fault system: a lithospheric-scale approach. In: HOLDSWORTH, R.E., STRACHAN, R.A. & DEWEY, J.F. (eds) Continental Transpressional and Transtensional Tectonics. Geological Society, London, Special Publications, 135, 143-158. VAN VEEN, G.W. 1966. Note on a JurassicCretaceous section in the Subbetic S W of Caravaca (prov. Murcia, SE Spain). Geologic en Mijnbouw, 45,391-397. VERA, J.A., GARCIA-HERNANDEZ, M., LOPEZGARRIDO, A.C., COMAS, M.C., RUIZ-ORTIZ, P.A. & MARTIN-ALGARRA, A. 1982. La cordillera Betica. In: GARCIA, A. (ed.) El Cretdcico en Espana. Universidad Complutense de Madrid, Madrid, 515-632. VILAS, L, DABRIO, C.J., PELAEZ, J.R. & GRACIAHERNANDEZ, M. 2001. Dominios sedimentarios generados durante el periodo extensional Cretacico Inferior sobre Cazorla y Hellin (Beticas externas). Su implicacion en la estructura actual. Revista de la Sociedad Geologica de Espana, 14, 113-122. VlLLALAIN, J.J., OSETE, M.L., VEGAS, P. & GARCIA-

DUENAS, V. 1992. Nuevos resultados paleomagneticos en el Subbetico Interne; implicaciones tectonicas. Actas de las Sesiones Cientificas; HI Congreso Geologico de Espana, 1, 308-312. WITTINK, R.J. 1975. A note on an Upper Jurassic to Miocene section in the Subbetic north of Velez Blanco (Province of Almeria, SE Spain). GUA Papers of Geology, 7, 89-101. Yu, G., WESNOUSKY, S.G. & EKSTROM, G. 1993. Slip partitioning along major convergent plate boundaries. Pure and Applied Geophysics, 140, 183-210.

This page intentionally left blank

Flow patterns during exhumation of the Sambagawa metamorphic rocks, SW Japan, caused by brittle-ductile, arc-parallel extension TORU TAKESHITA1 & KOSHI YAGI2 Department of Earth and Planetary Systems Science, Graduate School of Science, Hiroshima University, Higashi-Hiroshima 739-8526, Japan (e-mail: toru @ letitbe.geol. sci. hiroshima-u. ac.jp) 2 Hiruzen Institute for Geology and Chronology, 161-1 Sai, Okayama 703-8248, Japan 1

Abstract: Mesoscopic and microscopic structural analyses of the high-pressure/temperature Sambagawa metamorphic rocks (accretion complexes), SW Japan, have been carried out. Deformation characterized by extreme layer-normal thinning and nearly arc-parallel stretching occurred during exhumation in the Late Cretaceous. Asymmetric quartz c-axis fabrics and orientation of shear band cleavages reveal a pervasive top-to-the-west sense of shear in the Sambagawa metamorphic rocks during the exhumation stage. The 3D strain geometries, inferred from quartz c-axis fabric patterns, vary from plane strain to flattening across the metamorphic belt. We hypothesize that the data are most reasonably explained by a model of counter-flow in the subduction channel. The counter-flow was induced by a left-lateral oblique subduction of the oceanic (Izanagi) plate, which was strongly coupled with the subducting sediments. The 3D strain geometries suggest that the counter-flow (i.e. simple shear in the model) must have been accompanied by some arcnormal 'press' component. The mode of deformation changed from ductile to brittle arc-parallel extension, when the rocks were elevated and cooled below the temperature condition for the brittle-ductile transition of quartz (c. 300 °C). The normal faulting (i.e. brittle extension) at subgreenschist conditions was often accompanied by the precipitation of actinolite. This change in deformation mechanism with decreasing temperature is recorded by a conjugate set of normal faults found in the oligoclase-biotite zone in the study area, for which the palaeostress directions conform to the ductile strain geometries.

Exhumation mechanisms for high-P/T ratio type metamorphic rocks (orogens) under the setting of overall convergent plate (or plate subduction) boundaries have been debated. Among possible mechanisms (Fig. 1) are: (1) uplift and erosion of accreted sediments (e.g. Huerta et al 1999); (2) dynamic equilibrium between crustal thickening at depth (underplating) and extension at shallow level (Platt 1986); (3) strain partitioning caused by oblique plate subduction, leading to arc-parallel extension (Ave Lallemant & Guth 1990; McCaffrey 1992); (4) counter-flow in an accretionary wedge (e.g. Cloos 1982; Iwamori 2002); and (5) either buoyancy- or transpression-driven extrusion of subducted continent or sediments (e.g. Chemenda et al. 1995; Maruyama et al. 1996; Wallis 1998; Wintsch et al. 1999). In the models of (2) and (3), lateral extension plays a major role in the exhumation of orogens. Note also that in the models of (4) and (5), deformation during the exhumation

results in extreme layer-normal thinning and layer-parallel stretching, which is not easily distinguished from structures caused by lateral extension of orogens in the models (2) and (3). Because of the similarity of structures, the identification of the exhumation mechanism for high-P/r ratio type metamorphic rocks is difficult. In this chapter, we have documented finite strain geometries and sense of shear in the Sambagawa metamorphic rocks, central Shikoku, Japan, which resulted from ductile deformation during the exhumation stage. Furthermore, brittle deformation (normal faulting) in the Sambagawa metamorphic rocks, which occurred at the final exhumation stages, is evaluated for the first time. Based on these data, we discuss kinematics in the exhuming Sambagawa metamorphic rocks in relation to the oblique subduction of an oceanic plate, and mechanisms responsible for the exhumation.

From: GROCOTT, J., MCCAFFREY, K. J. W., TAYLOR, G. & TIKOFF, B. (eds) 2004. Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 279-296. 0305-8719/047$ 15 © The Geological Society of London 2004.

280

T. TAKESHITA & K. YAGI

Fig. 1. Five possible mechanisms of exhumation for high-P/J ratio type metamorphic rocks. In (1) erosion, the number indicates the order of underplating of sediments. In (3) strain partitioning, thick and thin arrows indicate the velocity vector of the relative motion of an oceanic plate, and arc-parallel components for the case of convex arc toward the ocean, respectively. Modified after McCaffrey (1992).

Geological outline The Sambagawa belt in SW Japan, a well-known high-P/r ratio type metamorphic belt, is juxtaposed in the north against the low-P/T ratio type Ryoke metamorphic belt, bounded by the Median Tectonic Line (MTL, Fig. 2). Both belts, which have been known as a paired metamorphic belt (Miyashiro 1961), originated from Jurassic accretionary complexes - see Faure et al (1991) for the Late Jurassic radiolarian fossils found in the Sambagawa metamorphic rocks. Minamishin et al. (1979) reported a Rb-Sr whole-rock isochron age of 116 ± 10 Ma for pelitic schist samples, interpreted as the age

of peak metamorphism (Isozaki & Itaya 1990). Amphibole and muscovite K-Ar (or 40Ar/39Ar) ages range from 94 to 65 Ma (e.g. Itaya & Takasugi 1988; Takasu & Dallmeyer 1990), probably indicating the ages of exhumation. On the basis of peak-metamorphic grade, the Sambagawa metamorphic rocks in central Shikoku can be divided into four zones based on the appearance of index minerals in pelitic schists: chlorite (300-360 °C, 5.5-6.5 kbar), garnet (440 ± 15 °C, 7-8.5 kbar), albite-biotite (520 ± 25 °C, 8-9.5 kbar) and oligoclasebiotite (610 ± 20 °C, 10-11 kbar) zones in order of increasing metamorphic grade (e.g. Higashino 1990; Enami et al 1994). It has been shown by

FLOW PATTERNS AND EXHUMATION, SW JAPAN

281

Fig. 2. (a) Simplified index map of SW Japan and Shikoku Island, (b) Metamorphic zonal map of the Sambagawa metamorphic belt, central Shikoku (study area). After Higashino (1990). (c) Cross-section across central Shikoku (Kamio-Asemi River section). Modified after Banno et al. (1978, 1986), Takasu & Dallmeyer (1990) and Takasu et al. (1994). The traverse line (I-F) of the cross-section (c) is shown in (b). MTL, Median Tectonic Line; I-STL, Itoigawa-Shizuoka Tectonic Line.

thermodynamic calculations that garnet rather than biotite appears first with increasing temperature under higher-pressure conditions than c. 5 kbar (Tinkham et al. 2001). In central Shikoku, metamorphic zonal mapping shows a peculiar structure, where the highest-grade oligoclase-biotite zone is located in the middle of the structural sequence (Fig. 2). This structure has been interpreted as a large-scale recumbent fold (Banno et al 1978; Wallis et al 1992) or as thrust sheets (Kara et al 1977; Faure 1985; Higashino 1990) formed after the peak metamorphism. It should be noted that a tectonic belt (or terrane) called the palaeo-Ryoke belt tectonically

overlays the Sambagawa belt, although this belt is only sporadically distributed along the MTL except for central Kyushu in SW Japan (Fig. 3; Takagi & Shibata 2000; Sakashima et al 2003). The palaeo-Ryoke belt consists of a variety of Palaeozoic-Mesozoic continentalshelf sedimentary and plutonic rocks (e.g. Takagi & Shibata 2000), and hence is inferred to have been a long-lived continental terrane. The continental-shelf sedimentary rocks suffered various degrees of metamorphism up to amphibolite to granulite facies, which belong to low- to intermediate-P/r ratio type (e.g. Takeda et al 1993; Obata et al 1994). Pressure at

282

T. TAKESHITA & K. YAGI

Fig. 3. Simplified geological map of western Shikoku and central Kyushu. Modified after Geological Survey of Japan (1992) and Takagi & Shibata (1996). For the mapped area, see Figure 2a. For the small distribution of the palaeo-Ryoke belt and Cretaceous sedimentary rocks in Oshima and surrounding area, see Takeda et al (1993) and Sakashima et al. (2003). 1, Upper Cretaceous sedimentary rocks (Onogawa, Tano and Mifune Groups); 2, Palaeozoic to Mesozoic continental-shelf metasediments and plutonic rocks (correlated with the palaeo-Ryoke belt); 3, Fault and inferred fault; MTL, Median Tectonic Line; OKTL, Oita-Kumamoto Tectonic Line (extension of the MTL); BTL, Butsuzo Tectonic Line.

peak-metamorphic temperature is inferred to be c. 5 kbar based on chemical compositions of amphibole in epidote-amphibolite samples (Takeda et al 1993). Recently, Sakashima et al (2003) have reported new SHRIMP zircon U-Pb ages of c. 110 Ma (late Early Cretaceous) for a group of plutonic rocks (or orthogneisses) and a paragneiss from the palaeo-Ryoke belt. These dates are comparable with the inferred age for peak metamorphism of the Sambagawa metamorphic rocks. These data suggest that the Sambagawa belt represents an Early Cretaceous subduction zone with the palaeo-Ryoke belt being the overlying continental crust. Since there is a pressure gap of peak metamorphism between the Sambagawa and palaeo-Ryoke belts, it is inferred that the structural discontinuity (fault) that bounds these two belts is a normal fault, which formed during the exhumation stage (Wheeler & Butler 1994). In fact, these two belts are separated by a low-angle normal fault at one locality in the Kanto mountains (Kobayashi 1996). Deformation structures, which we observe in outcrops and thin sections, mostly formed under retrograde conditions during the exhumation stages. They formed in three distinct phases, DI, D2 and D3 (e.g. Hara et al 1977; Faure 1983; Wallis 1990). The DI is characterized by a penetrative east-west-trending subhorizontal schistosity and lineation formed by ductile flow. The shape preferred orientation of amphiboles defines the lineation and formed under retrograde conditions during the exhumation stages (Hara

et al 1990, 1992; Nakamura & Enami 1994; Wintsch et al 1999; Banno 2000; Yagi & Takeshita 2002). Also, both Tagami & Takeshita (1998) and Yagi & Takeshita (2002) have shown that quartz c-axis fabrics formed at lower temperatures than those for the peak metamorphism based on their inferred glide systems. These conclusions indicate that the Sambagawa metamorphic rocks suffered an intense, penetrative deformation during the exhumation stages. D2 is recognized by south-vergent overturned folds with crenulation cleavages. It is dominant in the chlorite zone. D3 resulted in the formation of open upright folds with horizontal east-west trending axes, the wavelength of which varies from microscopic (thin section) to macroscopic (geological map) scale (e.g. Takeshita & Hara 1998). The D3-stage folds are considered to have formed after substantial exhumation (Banno & Sakai 1989), in relation to left-lateral displacement along the MTL (Hara et al 1977; Shiota et al 1993).

Results Mesoscopic structures The orientations of foliations and lineations have been measured extensively in the study area (Fig. 4). Foliations are defined by the shape preferred orientation of phengite; the lineations formed by either elongated quartz or the shape preferred orientation of amphibole. The geological structure in the study area is controlled by the

FLOW PATTERNS AND EXHUMATION, SW JAPAN

283

Fig. 4. Geological map of mesoscopic structures (foliation and mineral or stretching lineation). Equal-area and lower-hemisphere projections of poles to foliation and lineation are also shown. In the projection of poles to foliation, the best-fit great circle is drawn, the pole to which indicates the average orientation of the D3-stage fold axes. The projections of lineation are divided into two plots, each of which represents those for the east-west-trending (most of the study area) and NE-SW-trending (only the northern uppermost garnet and albite-biotite zones) lineation zones, respectively.

large-scale D3-stage folds (Tsuneyama synform and Yakushi antiform). In the southern part of the Tsuneyama synform, foliation strikes east-west to WNW-ESE and dips north to

NNE at moderate angles (20-60°). In the northern part of the Tsuneyama synform and southern part of the Yakushi antiform, foliation strikes east-west to WNW-ESE and dips south to SSE

284

T. TAKESHITA & K. YAGI

at moderate angles (30-60°). Note however that the foliation locally dips at high angles (>60°) in the northern upper chlorite, garnet and albite-biotite zones (Dozan, Saruta and Kamio Rivers, hereafter referred to as the Saruta River area, Fig. 4). In an axial part of the Yakushi antiform, the foliation dips at low angles (<30°). Further north of the Yakushi antiform, the foliation again strikes east-west to WNW-ESE and dips north to NNE at moderate to high angles (40-90°), and D3-stage folds with short wavelength develop near the MTL. The equalarea, lower-hemisphere projections of poles to the foliation reflect the D3-stage fold structure in the study area, which lie on a great circle (Fig. 4). The average orientation of the Ds-stage fold axes is determined to be trending N76°W and horizontal based on the projection. Parasitic folds associated with the large-scale D3-stage folds, the wavelength of which is less than a few metres, develop mostly in the chlorite zone and near the Tsuneyama synform and Yakushi antiform. Both amphibole and quartz stretching lineations are fairly uniform and trend east-west to WNW-ESE. The lineations are horizontal or

plunge either east or west at low angles (Fig. 4). Note here that the lineations in the northern uppermost garnet and albite-biotite zones trend NE-SW and plunge SW at moderate to high angles (i.e. almost oriented in the down-dip directions of the foliation), nearly orthogonal to the general direction of lineation in this area (Fig. 4). 3D strain geometries In quartz schist from the Sambagawa metamorphic belt in central Shikoku, 3D strain geometries resulting from the Dt-stage deformation were estimated, based on the symmetry of quartz c-axis fabric patterns (Tagami & Takeshita 1998). Small circle girdle c-axis fabrics about the Z-axis, and those about the X-axis (here referred to as cleft girdle c-axis fabrics), both of which exhibit axial symmetry (Fig. 5a), indicate uniaxial shortening (k = 0, A;-values after Flinn 1962) and elongation (k = oo), respectively (e.g. Lister & Hobbs 1980). Here, the X- and Z-axes denote the long and short axes of finite strain, and are assumed to be parallel to the lineation and normal to the foliation, respectively. Crossed girdle c-axis

Fig. 5. (a) Four c-axis fabric patterns, small circle, type I crossed, cleft and type II crossed girdles, in quartz schist from the study area, each of which is further divided into symmetric and asymmetric types. Both a natural example and a skeleton are shown for each fabric pattern. Upper-hemisphere and equal-area projection. Contour intervals are 1%, 2% and 3% per 1% area and shaded above 1% per 1% area. The determination of sense of shear for asymmetric girdles (indicated by arrows) is based on the disposition of the leading (solid line) and trailing (dashed line) edges of quartz c-axis fabric skeletons (Law 1987). (Continued)

FLOW PATTERNS AND EXHUMATION, SW JAPAN

285

Fig. 5. (Continued) (b) The spatial distribution of the quartz c-axis fabric patterns indicated by different symbols (inset). Compiled after Tagami (1998), Tagami & Takeshita (1998), Yagi & Takeshita (2002) and K. Okumura, N. Takao, A. Yoshida and N. Yamaguchi (unpublished data), (c) Projection of the inferred 3D strain geometries on the cross-section (Fig. 2c). Long thin and short thick dashed lines denote the metamorphic zone boundaries (see Fig. 2c) and those for the 3D strain geometries, respectively.

286

T. TAKESHITA & K. YAGI

fabrics show an orthorhombic symmetry that indicates pure shear (plane strain, k= 1). Since it is difficult to infer the exact fc-value for each c-axis fabric pattern, we group them into three patterns, approximately small circle, crossed and cleft girdles. These three patterns correspond to flattening (k = 0-1), plane strain (k = 1) and constriction (k = l-oo), respectively. The quartz c-axis fabrics often show monoclinic symmetry (180° rotation symmetry) about the F-axis, indicating a shear component parallel to the lineation (described below). However, even if the c-axis fabrics show monoclinic symmetry (or asymmetric patterns), the overall finite strain geometries can still be recognized by their patterns. We show asymmetric small circle, crossed (including single) and cleft girdle c-axis fabrics in Figure 5a, which were probably formed, respectively, by the combination of coaxial flattening, plane strain and constriction with a simple shear component. Note that the crossed girdle quartz c-axis fabrics in quartz schist from the entire study area mostly belong to type I (Lister 1977). Type II crossed girdle c-axis fabric (Lister 1977) only occurs along the Asemi River in the uppermost oligoclase-biotite zone and lower part of the albite-biotite zone of the upper structural level (Fig. 5b). The c-axis fabric transition from type I to type II crossed girdles was caused by increasing deformation temperatures at c. 450 °C (e.g. Takeshita 1996), and its tectonic implication is discussed elsewhere (Yagi & Takeshita 2002). The spatial distributions of the c-axis fabric patterns, symmetric and asymmetric small circle, type I crossed, cleft and type II crossed girdles, and inferred 3D strain geometries are shown in Figure 5b and c. The spatial distributions of the zones characterized by c-axis fabric patterns (i.e. 3D strain geometries) could be extended laterally, and are summarized below in order of increasing structural level in the Besshi nappe (Fig. 5c). Both cleft and type I crossed girdles are dominant in the upper chlorite zone. The upper chlorite zone along the Asemi River can be further divided into a lower cleft (constriction) and upper type I crossed (plane strain) girdle dominant zones. Both small circle (flattening) and type I crossed (plane strain) girdles are dominant in the garnet zone and the albite-biotite zone of the lower structural level. Either type I or II crossed girdles (plane strain) are dominant in the oligoclase-biotite zone and the albite-biotite zone of the upper structural level. Finally, small circle girdles (flattening) are dominant in the garnet zone of the upper structural level.

Sense of shear The sense of shear can be inferred from a variety of microstructures in the Sambagawa metamorphic rocks, such as asymmetric quartz c-axis fabric, shape preferred orientation of recrystallized quartz grains (Sq) oblique to the main schistosity (Sm) (oblique foliation, Passchier & Trouw 1996), shear band cleavages, etc. The sense of shear inferred from either/both the asymmetry of quartz c-axis fabrics or/and asymmetric disposition of shear band cleavages, discussed in this chapter, is completely consistent with those inferred from the other asymmetric structures (Fig. 6).

Fig. 6. Inferred sense of shear from different asymmetric microstructures in a quartz schist, (a) Foliation defined by the shape fabric of recrystallized quartz grains (5q) oblique to that defined by the preferred alignment of phengite (5m). (b) Quartz c-axis fabric showing monoclinic symmetry (almost a single girdle pattern). Upper-hemisphere and equal-area projection. Contour intervals are 1%, 2%, 3% and 4% per 1% area and shaded above 1 % per 1 % area. The sample locality is denoted in Figure 5b. Reproduced after Yagi & Takeshita (2002).

FLOW PATTERNS AND EXHUMATION, SW JAPAN

The inference of sense of shear from asymmetric oaxis fabrics is based on the internal asymmetry. In particular, our determination of sense of shear for crossed girdle type oaxis fabrics is based on the recognition of the leading and trailing edges of oaxis fabric skeleton (Fig. 5a; Law 1987). The sense of shear is

287

the same as that of rotation from the latter to the former about the Y-axis. The spatial distribution of the inferred sense of shear, which is indicated by an arrow with the head pointing toward the movement direction of the upper plate, is shown in Figure 7. It is evident from Figure 7 that the sense of shear is

Fig. 7. Diagram showing the spatial distribution of the sense of shear inferred from either/both asymmetric quartz oaxis fabric in quartz schist or/and orientation of shear band cleavages in quartz or basic schists.

288

T. TAKESHITA & K. YAGI

exclusively top-to-the-west in the southern part (i.e. Asemi and Shimokawa Rivers). A topto-the-east sense of shear rarely occurs in the lowermost structural level of the upper chlorite zone along the Shimokawa River, and in the middle of the oligoclase-biotite zone along the Asemi River. However, the reverse sense of shear is clearly associated with a late-stage overturned fold in these areas. In the northern part, the situation is more complex. A top-to-the-east sense of shear dominates in the upper chlorite and garnet zones, which is in contrast to the sense of shear in the southern part. In the albite-biotite zone, where the lineation direction trends NE-SW (nearly down-dip direction) as described above, a topto-the-NE sense of shear alternates with a topto-the-SW sense of shear. In the northernmost part of the study area near the MTL, a top-to-the west sense of shear again dominates. Brittle deformation (normal faulting) Although faults of various scales develop throughout the Sambagawa metamorphic belt in central Shikoku, late-stage faulting and folding particularly affected strata in the oligoclasebiotite zone, and the Saruta River area. The fault zones are often marked by the precipitation (or crystallization) of secondary minerals such as actinolite. Locally, the precipitation of actinolite is directly associated with the occurrence of serpentinite (Mount Shiraga and Dozan River, Fig. 2b; cf. Moore 1984). In the oligoclase-biotite zone north of Mount Shiraga, minor faults are abundant, with a breccia zone less than a few centimetres wide and a normal displacement less than a few metres (Fig. 8a, b). Consolidated breccia zones with a width of a few tens of centimetres also occur (Fig. 8c). We have analysed the palaeostress field for the minor normal fault systems using the inversion method of Angelier (1979). There are two sets of minor normal faults with striations: one set trending NE-SW to east-west and dipping NW to north at moderate angles, and the other set trending north-south to NNWSSE and dipping east at moderate angles (Figs 8a, b & 9a). These two sets of normal faults seem to be conjugate at outcrops (the former set either cuts the latter, or is cut by the latter, Fig. 8a, b). Although the strikes of the two sets of faults are significantly different, the striation directions are fairly similar, trending WNWESE to NW-SE (Fig. 9a). The striations indicate that the first set of normal faults has a left-lateral strike-slip component, while the second set has a right-lateral strike-slip component. The inversion

of the fault data with the Angelier (1979) method indicates that the compression and tension axes are nearly normal to the foliation and parallel to the lineation in this area, respectively (Figs 4 & 9b). The conclusion implies that the normal faults formed under the same stress field as caused the ductile deformation, after the rocks were elevated above the depth of brittleductile transition. On a geological map scale, NE-SW-trending faults with apparent left-lateral displacements ranging a few hundred metres to 1 km also develop in the oligoclase-biotite zone (Fig. 10). This set of geological map scale faults is probably correlated with the NE-SW- to east-west-trending minor normal faults with a left-lateral strike-slip component (Fig. 9a). Late-stage faulting and folding are extensive and large scale in the Saruta River area. We have measured recrystallized 'grain size' (diameter of equivalent circle to the XZ-section, average of 200-400 measurements) of quartz from quartz schist throughout the study area (Fig. 11 a). While the grain size increases continuously with increasing structural level in the southern part (Asemi River area), this general trend is disturbed in the northern part (Saruta River area), as shown by a few gaps in the recrystallized grain size. In fact, shear zones a few tens of centimetres wide occur at the gaps, which mostly consist of preferentially oriented fine-grained actinolite with lesser amount of chlorite, phengite, quartz, etc., suggesting that the faults were active under subgreenschist facies conditions (c. 300 °C). Furthermore, the spatial grain size distribution is clearly reversed (i.e. decreasing grain size with increasing structural level) in the northern albite-biotite zone, indicating late-stage overturned folding (Fig. 11). Discussion We discuss below if the structural features observed in the Sambagawa metamorphic rocks, which formed during the exhumation stages, are favourable or unfavourable for different models of exhumation (Fig. 1). Extreme ductile layer-normal thinning and arc-parallel stretching The development of the strong quartz c-axis fabrics indicates that a layer-normal shortening, >2030%, occurred during the exhumation stage (e.g. Takeshita & Wenk 1988). The large-strain ductile deformation during the exhumation can be caused by all the exhumation mechanisms except for erosion (model 1). However, both

FLOW PATTERNS AND EXHUMATION, SW JAPAN

Fig. 8. Development of normal faults in the oligoclase-biotite zone, (a) A conjugate set of minor normal faults in pelitic schist, (b) Sketch of (a), with arrows indicating the sense of displacement. Note that two types of quartz veins develop, which intruded parallel and oblique to the foliation, respectively, (c) 'Low-angle normal(?) fault' with a wide breccia zone almost parallel to the foliation. For localities of the outcrops of (a) and (c), see Figure 2b.

289

290

T. TAKESHITA & K. YAGI

Fig. 9. (a) Stereographic and lower-hemisphere projections of minor normal faults (great circles) and striations (indicated by an arrow on each fault plane), (b) Solution of the compression (P) and tension (7) axes for the given set of fault data in (a), using the Angelier (1979) method (calculated with the computer program FaultKin by Allmendinger et al. 1989). Fault data were collected in the area shown in Figure 2b.

gravity-driven extension (model 2) and buoyancydriven extrusion (model 5) could cause arc-normal extension, inconsistent with the arc-parallel stretching lineation. It must be noted that the plate tectonic setting during the exhumation of the Sambagawa metamorphic rocks was a strongly left-lateral oblique subduction of an oceanic plate underneath a continental plate (discussed below). Hence, the relative motion vector of the oceanic plate could have been decomposed into arcnormal and arc-parallel components (strain partitioning; Fitch 1972). In this situation, it has been shown that arc-parallel extension (or stretching) can occur for a convex arc toward the ocean (model 3; Ave Lallmant & Guth 1990; McCaffrey 1992). A nearly arc-parallel stretching can also be

caused by counter-flow in an accretionary wedge (model 4) during strongly oblique subduction of an oceanic plate, as discussed below. Uniform top-to-the-west sense of shear and its reversal by late-stage folding and faulting A top-to-the-east sense of shear in the northern upper chlorite and garnet zones, and both normal and reverse senses of shear in the northern albite-biotite zone (Fig. 7), appear to be difficult to explain. However, these zones are structurally disturbed as shown by the steep foliations, down-dip lineations (although this observation is

Fig. 10. Geological map scale NE-SW-trending faults, which apparently show left-lateral strike-slip displacement (denoted by arrows). For the mapped area, see Figure 2b.

FLOW PATTERNS AND EXHUMATION, SW JAPAN

291

Fig. 11. (a) Spatial distribution of recrystallized grain size of quartz from quartz schist, which is projected on the I-I' cross-section (Fig. 2c). Recrystallized grain size is measured for some of the samples shown in Figure 5b. No measurements are shown for quartz schist from the northernmost and southernmost parts, (b) Interpretative cross-section along the I-I' traverse. An arrow indicates the slip direction along the fault located at the base of the albite-biotite zone. See text for detailed explanations.

mostly confined to the northern albite-biotite zone) and occurrences of overturned folds and shear zones. Furthermore, Yagi & Takeshita (2002) have shown that a normal fault zone, which formed during the exhumation stages, is situated in the northern albite-biotite zone, based on the change in retrograde compositional zoning of amphibole within a short distance (c. I km). Therefore, the normal and reverse senses of shear in the northern part (Saruta River area), which are closely associated with the structural disturbance, are perhaps attributed to an overturned structure (possibly imbricate structure consisting of many overturned limbs; Fig. 1 Ib). The fault movement occurred at lowtemperature conditions, below the brittle-ductile transition temperatures for quartz, c. 300 °C (Stockhert et al 1999; Stipp et al 2002). The uniform top-to-the-west sense of shear during the exhumation conforms to counterflow in an accretionary wedge (model 4) under the setting of left-lateral oblique subduction of

an oceanic plate, but not extrusion (model 5), as discussed below. For extrusion of subducted sediments, the sense of shear must be reversed across the extruded edge (or centre) (e.g. Robin & Cruden 1994), which has not been observed in the study area. Wallis et al. (1992) and Wallis (1995) also inferred that a top-to-the-west sense of shear dominates in the Besshi nappe, but reported a few occurrences of the reverse sense of shear in the underlying Oboke nappe, based on various kinematic indicators. Wallis (1998) suggested that an extrusion of the Besshi nappe is preferred to a counter-flow model, based on the large-scale Besshi-Oboke tectonic discontinuity, although the reversal of sense of shear mentioned above is not sufficient to distinguish the two types of model. The uniform sense of shear might also fit to strain partitioning (model 3), if the arc-parallel extension was mostly accommodated by simple shearing, as is the case for the Cordillera core complexes (e.g. Lister & Davis 1989).

292

T. TAKESHITA & K. YAGI

Variable strain geometry of the exhuming rocks The strain geometries caused by ductile flow during part of the exhumation stages are not uniform, but rather vary from flattening to plane strain (Fig. 5). In the model of transpression (i.e. extrusion), where a deforming zone (viscous body) is sandwiched between two rigid plates, the strain geometry can change across the zone from plane strain at the boundary to flattening at the edge (or centre) of extrusion (e.g. Robin & Cruden 1994; Button 1997). On the other hand, only simple shear (plane strain) can be generated by counter-flow in a subduction channel (model 4). Hence, we can attribute the origin of the variable strain geometries from flattening to plane strain to transpression (model 5), or at least some arc-normal 'press' component in the counter-flow. The constrictive strain geometries, which occur in the southern upper chlorite zone, are anomalous. Takasu & Dallmeyer (1990) inferred that the upper chlorite, garnet and biotite zones (called Besshi nappe), and the lower chlorite zone (called Oboke nappe), are bounded by a fault (Fig. 5c), based on a gap of K-Ar ages. The constrictive zone lies directly on the inferred nappe boundary, suggesting the genetic relationship between the formation of the constrictive zone and the nappe emplacement. It is probable that the high-grade Besshi nappe, which was originally situated below the low-grade Oboke nappe, could have been emplaced on the Oboke nappe by counter-flow (model 4). If this is the case, a down-flow followed by a counter-flow (i.e. corner-flow) under the setting of oblique plate subduction could have produced the constrictive strain ellipsoids at the base of the Besshi nappe, as discussed by Iwamori (2002). Toriumi (1985) and Toriumi & Noda (1986) inferred a dominant prolate strain ellipsoid up to an axial ratio of 20 in the Besshi nappe, based on the deformed shape of radiolarian fossils, which recorded the entire deformation history from subduction to exhumation. These strain data are in contrast to the plane to flattening strain geometries inferred from the quartz c-axis fabric patterns, representing ductile flow during part of the exhumation stage. The prolate strain ellipsoid could support a corner-flow model, as discussed above.

A scenario for the exhumation of the Sambagawa metamorphic rocks Since proto-SW Japan (part of the Eurasian plate) was rotated clockwise by 50° during the opening

of the Japan Sea at c. 15 Ma (e.g. Otofuji & Matsuda 1984), it is inferred to have trended NNE-SSW in the Cretaceous (Fig. 12a). The exhumation of the Sambagawa metamorphic rocks of central Shikoku commenced after the peak metamorphism at c. 115 Ma and had progressed significantly by 85-75 Ma (muscovite K-Ar ages). The oceanic (Izanagi) plate is inferred to have been moving in the N25°W direction with respect to the Eurasian plate during these periods (Engebretson et al 1985; Maruyama & Seno 1986). Therefore, the velocity vector of the relative motion of the Izanagi plate made a 45° angle (rotated counterclockwise) with respect to the trend of proto-SW Japan (Fig. 12a). Under this tectonic setting of leftlateral oblique plate subduction, the uniform top-to-the-west sense of shear in the Sambagawa metamorphic rocks could be most reasonably explained by counter-flow in a subduction wedge (model 4), as explained below. In a corner-flow model, it is assumed that the accretionary prism is coupled both with the down-going oceanic plate at the bottom boundary (MO = MP, v0 = vp in Fig. 12b, c) and with the upper plate (continental crust) at the top boundary (ut = 0, vt = 0 in Fig. 12b, c). Note that a strong coupling at the bottom boundary is only required for the subduction and exhumation of an accretionary prism to occur. Although either a no-slip or a no-shear condition can be applied to the top boundary, the former is preferred here to explain upper crustal extension (discussed below). Under these boundary conditions, the arc-normal (u) and arc-parallel (v) velocity vectors as functions of the vertical distance (height) from the down-going oceanic plate are shown schematically in Figures 12b and c (e.g. Otsuki 1992). The total velocity vector in the accretionary prism, which is the combination of the arc-normal and arc-parallel components of the velocity vector, decreases in magnitude and rotates clockwise with increasing height (helical flow, Fig. 12d). Above the level of u = 0, where the velocity vector is parallel to the direction of the arc, the flow in the accretionary prism is converted from down- to counter-flow. In the model of corner-flow during left-lateral oblique plate subduction, the sense of shear is uniformly top-to-the-west (SSW before the opening of the Japan Sea), because the total velocity vector, which is always oriented eastward, continuously decreases with the increasing height in the subduction channel (Otsuki 1992, see Fig. 12d). This fact explains the uniform top-to-the-west sense of shear observed throughout the study area. Furthermore, the inverted structure of peak-metamorphic conditions from

FLOW PATTERNS AND EXHUMATION, SW JAPAN

293

the chlorite to oligoclase-biotite zone in the lower half of the Besshi nappe could be explained by the formation of a large-scale recumbent fold (Banno et al. 1978). It is in fact generated at a corner region in the corner-flow model (Iwamori 2002). This type of large-scale recumbent fold is difficult to reconcile with strain partitioning (model 3). Another important point is that the counterflow resulting from the oblique subduction exerts a large drag force at the bottom of the upper plate (continental crust, Fig. 12c), and hence arc-parallel extension in the upper plate. In fact, the upper plate consisting of the palaeoRyoke belt also underwent a significant amount of brittle-ductile arc-parallel stretching (see Sakashima et al. 2003, and references therein). Extensional basins (e.g. Onogawa, Tano and Mifune Groups in Kyushu shown in Fig. 3, Teraoka 1970) formed in the upper plate in the Late Cretaceous, coeval with the exhumation of the Sambagawa metamorphic rocks, indicating upper crustal extension.

Conclusions Mesoscopic and microscopic structural analyses of the high-P/r Sambagawa metamorphic rocks (accretion complexes) indicate that pervasive, nearly arc-parallel stretching occurred during the main exhumation stage named D! stage. Although the deformation in the subduction channel was probably caused by a left-lateral oblique subduction of the oceanic (Izanagi) plate during the Late Cretaceous, the detailed mechanism for the exhumation is not clear. Nevertheless, a pervasive top-to-the-west sense of shear inferred from asymmetric quartz c-axis fabrics and orientation of shear band cleavages supports a model of cornerflow. Corner-flow requires that the subducting oceanic plate and sediments are strongly coupled. The 3D strain geometries, inferred from quartz c-axis fabric patterns, were caused by ductile flow during part of the exhumation stage, and

Fig. 12. Proposed corner-flow model for the exhumation of the Sambagawa metamorphic rocks, (a) Inferred configuration of the Eurasian continental and Izanagi oceanic plates, and relative motion vector of the Izanagi to Eurasian plate in the early Late Cretaceous (100-85 Ma). After Engebretson et al. (1985) and Maruyama & Seno (1986). (b, c) Schematic diagrams showing the calculated arc-normal and arc-parallel velocity components of

corner-flow in the accretionary prism (Sambagawa metamorphic rocks), respectively, resulting from the leftlateral oblique subduction of the oceanic plate in the early Late Cretaceous. Normal faults and extensional basins, which were generated as a result of drag force at the bottom of the continental crust (palaeo-Ryoke belt), are shown in (c). (d) Schematic diagram showing the net velocity vectors of corner-flow in accretionary prism with increasing vertical distance (height) from the subducting oceanic plate, for the case of left-lateral oblique subduction of an oceanic plate. Modified after Otsuki (1992). See text for details.

294

T. TAKESHITA & K. YAGI

mostly vary from plane strain to flattening. This suggests that the counter-flow (i.e. simple shear in the model) must have been accompanied by some arc-normal 'press' component. Anomalous constrictive strain ellipsoids occur at the base of the high-grade Besshi nappe. These structures could be explained by a corner-flow-induced nappe emplacement during oblique plate subduction, in which a down-flow followed by a counter-flow resulted in the constrictive strain geometries (Iwamori 2002). Finally, arc-parallel extension proceeded even after the rocks were exhumed above the depth of the brittle-ductile transition of quartz, and hence cooled below the temperature conditions of c. 300 °C. This is recorded by a conjugate set of normal faults found in the oligoclase-biotite zone in the study area, for which the palaeostress directions conform to the D! strain geometries. The normal faulting at subgreenschist conditions, which was often accompanied by the precipitation of actinolite, was extensive and caused overturned folding in some places, as shown by the reversal of the shear sense. The manuscript benefited from constructive reviews by H. Ave Lallemant and R. Wintsch. B. Tikoff is thanked for his editorial corrections of the manuscript and valuable comments. This study was supported in part by a grant from the Japan Society for the Promotion of Science (No. 14340151) to T. Takeshita.

References ALLMENDINGER, R.W., MARRETT, R.A. & CLADOUHOS, T. 1989. FaultKin 3.8a academic version; a program for analyzing fault slip data. Computer program. ANGELIER, J. 1979. Determination of the mean principal directions of stresses for a given fault population. Tectonophysics, 56, 17-26. AVE LALLEMANT, H.G. & GUTH, L.R. 1990. Role of extensional tectonics in exhumation of eclogites and blueschists in an oblique subduction setting: northeastern Venezuela. Geology, 18, 950-953. BANNO, S. & SAKAI, C. 1989. Geology and metamorphic evolution of the Sanbagawa metamorphic belt, Japan. In: DALY, J.S., CLIFF, R.A. & YARDLEY, B.W.D. (eds) Evolution of Metamorphic Belts. Geological Society, London, Special Publications, 43, 519-532. BANNO, S., HIGASHINO, T., OTSUKI, M., ITAYA, T. & NAKAJIMA, T. 1978. Thermal structure of the Sanbagawa metamorphic belt in central Shikoku. Journal of Physics of the Earth, 26, Supplement, 345-356. BANNO, S., SAKAI, C. & HIGASHINO, T. 1986. Pressure-temperature trajectory of the Sanbagawa metamorphism deduced from garnet zoning. Lithos, 19, 51-63.

BANNO, Y. 2000. Intermediate high-pressure exhumation of the northern segment of the Sanbagawa belt, Saruta-gawa area, central Shikoku, Japan. Lithos, 50, 289-303. CHEMENDA, A.I., MATTAUER, M., MALAVIEILLE, J. & BOKUN, A.N. 1995. A mechanism for syn-collisional rock exhumation and associated normal faulting: results from physical modelling. Earth and Planetary Science Letters, 132, 225-232. CLOOS, M. 1982. Flow melanges: numerical modeling and geologic constraints on their origin in the Franciscan subduction complex. Geological Society of America Bulletin, 93, 330-345. DUTTON, BJ. 1997. Finite strains in transpression zones with no boundary slip. Journal of Structural Geology, 19, 1189-1200. ENAMI, M., WALLIS, S.R. & BANNO, Y. 1994. Paragenesis of sodic pyroxene-bearing quartz schists: implications for P-T history of the Sanbagawa belt. Contributions to Mineralogy and Petrology, 116, 182-198. ENGEBRETSON, D.C., Cox, A. & GORDON, R.G. 1985. Relative motions between oceanic and continental plates in the Pacific basin. Geological Society of America, Special Paper, 206, 1-59. FAURE, M. 1983. Eastward ductile shear during the early tectonic phase in the Sanbagawa belt. Journal of the Geological Society of Japan, 89, 319-329. FAURE, M. 1985. Microtectonic evidence for eastward ductile shear in the Jurassic orogen of SW Japan. Journal of Structural Geology, 7, 175—186. FAURE, M., IWASAKI, M., ICHIKAWA, K. & YAO, A. 1991. The significance of Upper Jurassic radiolarians in high pressure metamorphic rocks of SW Japan. Journal of Southeast Asian Earth Sciences, 6, 131 — 136. FITCH, T.J. 1972. Plate convergence, transcurrent faults, and internal deformation adjacent to Southeast Asia and the western Pacific. Journal of Geophysical Research, 77, 4432-4460. FLINN, D. 1962. On folding during three dimensional progressive deformation. Geological Society of London, 118, 385-433. Geological Survey of Japan 1992. Geological Map of Japan, 1:1000000 (3rd edn.). HARA, I., HIDE, K., TAKEDA, K., TSUKUDA, E., TOKUDA, M. & SHIOTA, T. 1977. Tectonic movement in the Sambagawa belt. In: HIDE, K. (ed.) The Sambagawa Belt. Hiroshima University Press, Hiroshima, 307-390 (in Japanese with English abstract). HARA, I., SHIOTA, T., HIDE, K., OKAMOTO, K., TAKEDA, K., HAYASAKA, Y. & SAKURAI, Y. 1990. Nappe structure of the Sambagawa belt. Journal of Metamorphic Geology, 8, 441-456. HARA, I., SHIOTA, T. etal 1992. Tectonic evolution of the Sambagawa schists and its implications in convergent margin processes. Journal of Science of the Hiroshima University, Series C, 9, 495—595. HIGASHINO, T. 1990. The higher grade metamorphic zonation of the Sambagawa metamorphic belt in central Shikoku, Japan. Journal of Metamorphic Geology, 8, 413-423.

FLOW PATTERNS AND EXHUMATION, SW JAPAN HUERTA, A.D., ROYDEN, L.H. & HODGES, K.V. 1999. The effects of accretion, erosion and radiogenic heat on the metamorphic evolution of continental orogens. Journal of Metamorphic Geology, 17, 349-366. ISOZAKI, Y. & ITAYA, T. 1990. Chronology of Sanbagawa metamorphism. Journal of Metamorphic Geology, 8, 401-411. ITAYA, T. & TAKASUGI, H. 1988. Muscovite K-Ar ages of the Sanbagawa schists, Japan and argon depletion during cooling and deformation. Contributions to Mineralogy and Petrology, 100, 281-290. IWAMORI, H. 2002. Some remarks on deformation and P—T conditions of the Cretaceous regional metamorphic belts in southwest Japan: reply to comment by M. Brown on Thermal effects of ridge subduction and its implications for the origin of granitic batholith and paired metamorphic belts'. Earth and Planetary Science Letters, 199, 493-501. KOBAYASHI, K. 1996. Rotation of slip direction of the Atokura Nappe viewed from micro-structural analyses of brittle shear zones in the Sambagawa belt, Southwest Japan. Journal of Structural Geology, 18, 563-571. LAW, R.D. 1987. Heterogeneous deformation and quartz crystallographic fabric transitions: natural examples from the Moine Thrust zone at the Stack of Glencoul, northern Assynt. Journal of Structural Geology, 9, 819-833. LISTER, G.S. 1977. Discussion. Crossed-girdle c-axis fabrics in quartzites plastically deformed by plane strain and progressive simple shear. Tectonophysics, 39, 51-54. LISTER, G.S. & DAVIS, G. 1989. The origin of metamorphic core complexes and detachment faults formed during Tertiary continental extension in the northern Colorado River region, USA. Journal of Structural Geology, 11, 65-94. LISTER, G.S. & HOBBS, B.E. 1980. The simulation of fabric development during plastic deformation and its application to quartzite: the influence of deformation history. Journal of Structural Geology, 2, 355-370. MARUYAMA, S. & SENO, T. 1986. Orogeny and relative plate motions: example of the Japanese Islands. Tectonophysics, 127, 305-329. MARUYAMA, S., Liou, J.G. & TERABAYASHI, M. 1996. Blueschists and eclogites of the world and their exhumation. International Geology Review, 38, 485-594. MCCAFFREY, R. 1992. Oblique plate convergence, slip vectors, and forearc deformation. Journal of Geophysical Research, 97, 8905-8915. MINAMISHIN, M., YANAGI, T. & YAMAGUCHI, M. 1979. Rb-Sr whole rock age of the Sanbagawa metamorphic rocks in central Shikoku, Japan. In: YAMAGUCHI, M. (ed.) Isotope Geosciences of Japanese Islands. Report for Scientific Research from the MEJ, 334054, 68-71 (in Japanese). MIYASHIRO, A. 1961. Evolution of metamorphic belts. Journal of Petrology, 2, 277-331.

295

MOORE, D.E. 1984. Metamorphic history of a highgrade blueschist exotic block from the Franciscan Complex, California. Journal of Petrology, 25, 126-150. NAKAMURA, C. & ENAMI, M. 1994. Prograde amphiboles in hematite-bearing basic and quartz schists in the Sanbagawa belt, central Shikoku: relationship between metamorphic field gradient and P-T paths of individual rocks. Journal of Metamorphic Geology, 12, 841—852. OBATA, M., YOSHIMURA, Y., NAGAKAWA, K., ODAWARA, S. & OSANAI, Y. 1994. Crustal anatexis and melt migrations in the Higo metamorphic terrane, west-central Kyushu, Kumamoto, Japan. Lithos,32, 135-147. OTOFUJI, Y. & MATSUDA, T. 1984. Timing of rotational motion of Southwest Japan inferred from paleomagnetism. Earth and Planetary Science Letters, 70, 373-382. OTSUKI, K. 1992. Uplifting and shear deformation of high-P metamorphic belt by oblique subduction of plate: an investigation by quasi-3-dimensional lubricant model. Journal of the Geological Society of Japan, 98, 435-444 (in Japanese with English abstract). PASSCHIER, C.W. & TROUW, R.A.J. 1996. Microtectonics. Springer-Verlag, Berlin. PL ATT, J.P. 1986. Dynamics of orogenic wedges and the uplift of high-pressure metamorphic rocks. Geological Society of America Bulletin, 97, 1037-1053. ROBIN, P.-Y.F. & CRUDEN, A.C. 1994. Strain and vorticity patterns in ideally ductile transpression zones. Journal of Structural Geology, 16, 447-466. SAKASHIMA, T., TERAD A, K., TAKESHITA, T. & SANO, Y. 2003. Large-scale displacement along the Median Tectonic Line, Japan: evidence from SHRIMP zircon U-Pb dating of granites and gneisses from the South Kitakami and paleo-Ryoke belts. Journal of Asian Earth Sciences, 21, 1019-1039. SHIOTA, T., HARA, I., SEKI, S., IKEDA, Y., OKUDAIRA, T. & HIDE, K. 1993. Sinistral en echelon folding of the Sambagawa schists and its tectonic implication. Journal of Science of the Hiroshima University, Series C, 9, 671-683. STIPP, M., STUNITZ, H., HEILBRONNER, R. & SCHMID, S.M. 2002. The eastern Tonale fault zone: a 'natural laboratory' for crystal plastic deformation of quartz over a temperature range from 250 to 700 °C. Journal of Structural Geology, 24, 18611884. STOCKHERT, B., BRIX, M.R., KLEINSCHRODT, R., HURFORD, A.J. & WIRTH, R. 1999. Thermochronometry and microstructures of quartz - a comparison with experimental flow laws and predictions on the temperature of the brittle-plastic transition. Journal of Structural Geology, 21, 351-369. TAGAMI, M. 1998. c-Axis fabrics and microstructures in quartz schist from the Sambagawa metamorphic belt, central Shikoku, Japan. PhD thesis, Hiroshima University, Japan. TAGAMI, M. & TAKESHITA, T. 1998. c-Axis fabrics and microstructures in quartz schist from the Sam-

T. TAKESHITA & K. YAGI

296

bagawa metamorphic belt, central Shikoku, Japan. Journal of Structural Geology, 20, 1549-1568. TAKAGI, H. & SHIBATA, K. 1996. Restoration of the Paleo-Ryoke Terrane. In: SHIMAMOTO, T., HAYASAKA, Y., SHIOTA, T., ODA, M., TAKESHITA, T., YOKOYAMA, S. & OHOTOMO, Y. (eds) Tectonics and Metamorphism (The Hara Volume). SOBUN, Tokyo, 211-219 (in Japanese with English abstract). TAKAGI, H. & SHIBATA, K. 2000. Constituents of the Paleo-Ryoke Belt and restoration of the PaleoRyoke and Kurosegawa Terranes. Memoirs of the Geological Society of Japan, 56, 1-12 (in Japanese with English abstract). TAKASU, A. & DALLMEYER, R.D. 1990. 40Ar/39Ar mineral age constraints for the tectonothermal evolution of the Sambagawa metamorphic belt, central Shikoku, Japan: a Cretaceous accretionary prism. Tectonophysics, 185, 111-139. TAKASU, A., WALLIS,S.,BANNO,S.&DALLMEYER,R.D. 1994. Evolution of the Sambagawa metamorphic belt, Japan. Lithos, 28, 69-84. TAKEDA, K., MAKISAKA, S., ITAYA, T. &NISHIMURA, Y. 1993. The Maana belt in the Mikame area, western edge of Shikoku, southwest Japan: composition and geotectonic setting. Journal of the Geological Society of Japan, 99, 255-279 (in Japanese with English abstract). TAKESHITA, T. 1996. Estimate of the physical conditions for deformation based on oaxis fabric transitions in naturally deformed quartzite. Journal of the Geological Society of Japan, 102, 211-222 (in Japanese with English abstract). TAKESHITA, T. & KARA, I. 1998. c-Axis fabrics and microstructures in a recrystallized quartz vein deformed under fluid-rich greenschist conditions. Journal of Structural Geology, 20, 417-431. TAKESHITA, T. & WENK, H.-R. 1988. Plastic anisotropy and geometrical hardening in quartzites. Tectonophysics, 149, 345-361. TERAOKA, Y. 1970. The Cretaceous system in the Onogawa basin. Report of the Geological Society of Japan, 237, 1-84 (in Japanese with English abstract). TlNKHAM, D.K.,

ZULUAGA, C.A.

& STOWELL,

H.H.

2001. Metapelite phase equilibria modeling in

MnNCKFMASH: the effect of variable A12O3 and MgO/(MgO + FeO) on mineral stability. Geological Materials Research, 3, 1-42. TORIUMI, M. 1985. Two types of ductile deformation/ regional metamorphic belt. Tectonophysics, 113, 307-326. TORIUMI, M. & NODA, H. 1986. The origin of strain patterns resulting from contemporaneous deformation and metamorphism in the Sambagawa metamorphic belt. Journal of Metamorphic Geology, 4, 409-420. WALLIS, S.R. 1990. The timing of folding and stretching in the Sambagawa belt: the Asemigawa region, central Shikoku. Journal of the Geological Society of Japan, 96, 345-352. WALLIS, S.R. 1995. Vorticity analysis and recognition of ductile extension in the Sanbagawa belt, SW Japan. Journal of Structural Geology, 17, 1077-1093. WALLIS, S.R. 1998. Exhuming the Sanbagawa metamorphic belt: the importance of tectonic discontinuities. Journal of Metamorphic Geology, 16, 83-95. WALLIS, S.R., BANNO, S. & RADVANEC, M. 1992. Kinematics, structure and relationship to metamorphism of the east-west flow in the Sanbagawa Belt, southwest Japan. The Island Arc, 1, 176-185. WHEELER, J. & BUTLER, R.W.H. 1994. Criteria for identifying structures related to true crustal extension in orogens. Journal of Structural Geology, 16, 1023-1027. WINTSCH, R.P., BYRNE, T. & TORIUMI, M. 1999. Exhumation of the Sanbagawa blueschist belt, SW Japan, by lateral flow and extrusion: evidence from structural kinematics and retrograde P-T-t paths. In: RING, U., BRANDON, M.T., LISTER, G.S. & WILLETT, S.D. (eds) Exhumation Processes: Normal Faulting, Ductile Flow and Erosion. Geological Society, London, Special Publications, 154, 129-155. YAGI, K. & TAKESHITA, T. 2002. Regional variation in exhumation and strain rate of the high-pressure Sambagawa metamorphic rocks in central Shikoku, southwest Japan. Journal of Metamorphic Geology, 20, 633-647.

Transtensional deformation at the junction between the Okinawa trough back-arc basin and the SW Japan island arc OLIVIER FABBRI1, PATRICK MONIE2 & MARC FOURNIER3 EA 2642, Universite de Franche-Comte, 16 route de Gray, 25030 Besanqon Cedex, France (e-mail: olivier.fabbri @ univ-fcomte.fr) 2UMR 5573, Universite des Sciences et Techniques du Languedoc, 2 Place E. Bataillon, 34095 Montpellier Cedex, France 3 UMR 7072, Universite Pierre et Marie Curie, 4 Place Jussieu, 75252 Paris Cedex 05, France 1

Abstract: The Okinawa trough is a back-arc basin currently forming above the subducting Philippine Sea plate by crustal stretching of the Eurasia lithosphere. Existing geophysical investigations have revealed that the northern part of the Okinawa trough consists of a series of en echelon left-stepping grabens or half-grabens, some of which were formed in Miocene times and are presently inactive. In the Kyushu island, three major zones of extensional deformation are recognized. They are characterized by 20-40 km long sediment-filled basins lying on the hanging-wall side of N60°- to N80°-trending, north- or NW-dipping normal faults. These basins display an en echelon left-stepping arrangement. The ages of faulted rocks, the ages of graben-filling sediments and radiometric ages newly obtained on pseudotachylites associated with normal faulting indicate that extension started at 13 Ma at the southern end of Kyushu and migrated northwards in the middle part of Kyushu (Beppu Bay), where it is still active today. The continentward-dipping, listric geometry inferred in depth for the fault systems is consistent with that of faults imaged by existing seismic profiles obtained off-shore in the Okinawa trough and across the Beppu Bay. The spatial association between Miocene or younger normal faults and pre-Miocene regional low-angle thrust faults suggests the possibility of a reactivation of some of the thrust faults as low-angle detachment faults merging in depth into a mid-crustal partial attachment zone.

Oblique rifting at divergent plate boundaries or along intracontinental rifts is a common process that has been described in many areas (Illies 1977; Angelier et al 1981; Withjack & Jamison 1986, and references therein; Boccaletti etal 1987; Dauteuil & Brun 1993, 1996; Schreurs & Colletta 1998, and references therein; Schumacher 2002). It has also been extensively studied through the means of analogue experiments (Withjack & Jamison 1986; Tron & Brun 1991; Schreurs & Colletta 1998; Acocella et al 1999; Basile & Brun 1999; Mart & Dauteuil 2000). Oblique rifting within back-arc basins in the upper plate at convergent margins is a less common process. A notable example is provided by the Havre trough, currently opening behind the Kermadec trench and propagating southwards through the New Zealand arc in the Taupo Volcanic Zone (Caress 1991; Wright 1993; Benes & Scott 1996; Parson & Wright 1996; Wright et al 1996; Delteil et al 2002). The Okinawa trough, currently opening behind the Ryukyu trench

between SW Japan and Taiwan, is another example of oblique rifting. The aim of this chapter is to analyse the transtensional deformation linked with oblique rifting in the junction area where the rift propagates into the SW Japan arc. We also propose a cross-sectional model for the junction area in which the scattered nearsurface extensional systems merge at depth into a hypothetical mid-crustal partial attachment zone as defined by Tikoff et al (2002).

Geodynamical and geological outline present-day plate configuration and recent evolution The study area is located behind the zone of convergence between the subducting Philippine Sea plate and the overriding Eurasia plate. The direction of relative convergence is about N45°W and the rate of convergence is between 4 and

From: GROCOTT, J., MCCAFFREY, K. J. W., TAYLOR, G. & TIKOFF, B. (eds) 2004. Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 297-312. 0305-8719/047$ 15 © The Geological Society of London 2004.

O.PAEERIETAL

298 1

5 cm a (DeMets et al. 1990; Seno et al 1993; El-Fiky et al 1999). The Quaternary to presentday volcanic arc associated with the Philippine Sea plate subduction can be traced across Kyushu and the western part of Honshu (Fig. 1; Kamata 1998). Further south, it corresponds to an alignment of volcanic islands located behind the non-volcanic Ryukyu arc (Notsu et al. 1987). It can be discontinuously traced to Taiwan (Sibuet et al 1995, 1998). In Kyushu, an east-west alignment of volcanoes, possibly related to back-arc rifting (Nakamura et al 1985; Nakada & Kamata 1991), crosses the middle part of the island and is superimposed on the SW-NE volcanic arc related to the subduction. The present-day N45°W direction of relative convergence has remained steady for the past 1.5 to 5 Ma. Before 1.5 to 5 Ma, the direction of relative convergence was NNW-SSE, as indicated by various onland shortening structures in central Japan (Matsuda 1980; Seno & Maruyama 1984; Angelier & Huchon 1987) and also in Taiwan (Angelier et al 1986). The Okinawa trough back-arc basin, which extends from Kyushu to Taiwan, has been forming since Miocene times behind the nonvolcanic Ryukyu arc by crustal stretching and thinning of the Eurasia lithosphere (Lee et al 1980; Kimura 1985; Letouzey & Kimura 1985, 1986; Sibuet et al 1987, 1995, 1998; Park et al 1998). The crust underlying the Okinawa trough is everywhere continental in nature and its thickness decreases from 27-30 km in the northeastern part near Kyushu to 15-18 km in the southern part near Taiwan (Iwasaki et al 1990; Hirata et al 1991; Sibuet et al 1995). Intraplate earthquakes reflecting tensional stresses inside the Eurasia lithosphere are clearly distinguishable from deeper interplate earthquakes distributed along a N45°E-trending, NW-dipping Wadati-Benioff plane (Eguchi & Uyeda 1983; Kao & Chen 1991). Whereas interplate earthquakes mostly show compressional mechanisms with NW-SE-directed P axes, intraplate earthquakes in the crust or upper mantle underlying the Okinawa trough display either extensional or strike-slip focal mechanisms (Kao & Chen 1991; Fabbri & Fournier 1999; Fournier et al 2001). The strike-slip type mechanisms are more numerous in the northern Okinawa trough than elsewhere, and most nodal planes strike between north-south and N20°E for one family, and between N90°E and Nl 10°E for the other family. Global Positioning System (GPS) measurements have shown that the maximum rates of extension across the Okinawa trough are found

in the southernmost part, where they reach 4 cm a"1 (Imanishi et al 1996). Elsewhere, the rates are 1 cm a~ ! or less. Structure of the northern part of the Okinawa trough In the northern part of the Okinawa trough, water depths never exceed 1000 m. The geological structure has been studied with the help of seismic refraction and reflection, magnetic and gravity profiles (Aiba & Sekiya 1979; Nash 1979; Sibuet et al 1987, 1995, 1998; Iwasaki et al 1990). Seismic reflection profiles (Nash 1979) reveal that the northern Okinawa trough is not a unique graben but rather is composed of a series of basins and ridges that are hardly recognizable in the present-day bathymetry due to an extensive Quaternary sedimentary cover (Figs 1 & 2). Despite this difficulty, the following features can be mapped from the SE to the NW: (1) the non-volcanic Ryukyu arc represented by the Yakushima and Tanegashima islands; (2) the volcanic arc, a broad zone composed of deformed sedimentary rocks covered or intruded by Plio-Quaternary to present-day volcanic deposits, lava flows and sills; (3) the Tokara sub-basin, a series of half-grabens filled with at least 3000 m of Miocene-Pliocene sedimentary strata; (4) the Tokara ridge, locally emerging in the Koshiki islands; (5) the Goto sub-basin, inside which the thickness of the Miocene Pliocene sedimentary strata exceeds 5000 m; (6) the Goto ridge; and (7) the East China Sea continental shelf. Southwestwards, the Tokara and Goto sub-basins merge to form a single graben system, the central Okinawa trough. The basement of the ridges and basins is Middle Miocene in age or older. The basement rocks of the Koshiki islands consist of Cretaceous to Palaeogene sandstones and siltstones intruded by Middle Miocene granodiorite plutons. The basement rocks of the Goto islands consist of Miocene volcano-sedimentary strata intruded by Middle Miocene granodiorites or covered by Quaternary alkaline basalts. The oldest strata filling the various basins is of Pliocene or Miocene age (Nash 1979; Letouzey & Kimura 1985). Inception of normal faulting, subsidence and block tilting in the northern Okinawa trough is not accurately dated but likely took place during the Late Miocene. The youngest tilted strata are Pleistocene in age. Two regional N20°E-trending zones of westdipping normal faults can be mapped (Fig. 1). The Tokara Line fault zone, separating the volcanic arc from the Tokara sub-basin, is a major,

TRANSTENSION IN THE SW JAPAN ARC SYSTEM

299

Fig. 1. Simplified structural map of the northern part of the Okinawa trough, northern Ryukyu arc and Kyushu island. Partly after Research Group for Active Faults of Japan (1991). AKTL, Amami-Kagoshima Tectonic Line; BTL, Butsuzo Tectonic Line; MTL, Median Tectonic Line; NTL, Nobeoka Tectonic Line. Cross-sections A-A' and B-B' are shown on Figure 2. Cross-section C—C' is shown on Figure 8. Cross-section D-D' is shown on Figure 10.

Fig. 2. Cross-sections across the northern part of the Okinawa trough. Modified from Nash (1979), with permission from the Japanese Association for Petroleum Technology. Location is given on Figure 1. Vertical scale is two-way travel time in seconds.

TRANSTENSION IN THE SW JAPAN ARC SYSTEM

listric-shaped growth fault, which can be followed downward to more than 6000 m on seismic sections. Its northward extension is not precisely known. According to Nash (1979), it connects with the Butsuzo Tectonic Line (BTL, Fig. 1) in Kyushu. We retained the more north-south trend proposed by the Research Group for Active Faults of Japan (1991). The Amami-Kagoshima Line stretches between the non-volcanic Ryukyu arc and the volcanic arc. To the SW of the study area, it constitutes the eastern boundary of the Amami sub-basin, a small half-graben filled with about 2000 m of Pliocene sediments (Fig. 2, section A-A'). The Amami-Kagoshima Line cannot be imaged at depth on seismic profiles. Its downward geometry is thus unknown. It is still active today, at least at the latitude of Yakushima. Indeed, the anomalously high elevation of the Yakushima island, located immediately to the SE of the Amami-Kagoshima Line and culminating at 1935 m, the highest point in SW Japan, indicates a significant active uplift of the footwall block. According to Nash (1979), the AmamiKagoshima Line can be traced through the Kikai and Ata calderas until the southern tip of Kyushu. The N60°E-trending normal faults crossing the southern part of the Osumi peninsula likely abut against the more northerly trending Amami-Kagoshima Line between the Ata and Kikai calderas. The faults bounding the Tokara and Goto half-grabens and other minor subbasins have shorter extents than the Tokara or Amami-Kagoshima lines and are slightly oblique to them (Fig. 1). They strike between N40°E and N50°E, drawing an en echelon pattern that suggests formation under a right-lateral transtensional shear along a N20°E direction. The asymmetric half-graben structures with master faults dipping away from the trench, which are characteristic of the northern part of the trough, are also recognized in the central part, where a multichannel reflection profile shows a probable detachment fault located between the Okinawa trough and the Ryukyu arc, and dipping northwards (Park et al 1998). Outline of the geological structure of Kyushu The southern half of the Kyushu island, which belongs to the so-called 'Outer Zone' of SW Japan, consists of a stack of imbricate thrust sheets composed predominantly of Jurassic to Palaeogene clastic rocks (Murata 1981, 1987, 1991; Sakai & Kanmera 1981; Taira et al 1982; Ogawauchi & Iwamatsu 1986; Nishi

301

1988; Kimura et al. 1991; Saito et al. 1996). The otherwise well-delineated Median Tectonic Line (MTL) of SW Japan disappears in eastern Kyushu, south of the Beppu Bay, beneath widespread Plio-Quaternary continental volcanic deposits. Clastic rocks involved in the thrust sheets get younger when going southwards closer to the Pacific Ocean. Jurassic rocks constitute thrust sheets located to the north of the Butsuzo Tectonic Line (BTL), a major regional thrust. Cretaceous rocks are found between the BTL and the Nobeoka Tectonic Line (NTL), another major regional thrust. Palaeogene to Lower Miocene strata are distributed southwards or eastwards of the NTL. A series of granitic to granodioritic plutons are intruded into the thrust sheets and their ages are well constrained and clustered around 14 Ma (Oba 1977; Shibata 1978). These plutons were considered as 'posttectonic'. Although they undoubtedly post-date thrust sheet formation, several studies have shown that they are affected by normal faults (Oba 1961; Obata 1961; Nozawa & Ota 1967; Saito et al 1996; Fabbri et al 1997). Diffuse extension across south Kyushu: geometry and kinematics of faults and chronology In the southern half of Kyushu, post-15 Ma extensional faulting is observed in three key areas, which will be successively described. Osumi region The Osumi peninsula is located to the south of the Kyushu island (Fig. 1). Its southernmost tip is occupied by the Middle Miocene Osumi granodioritic pluton intruded in Eocene to Oligocene sandstones and mudstones of the Shimanto Group (Figs 3 & 4). The pluton is cut by a series of N60°E-trending normal faults dipping predominantly northwards (Obata 1961; Fabbri et al 1997). Fault plane dips are between 45° and 80°. Given the lack of Miocene syntectonic deposits overlapping the faulted granodiorite, it is impossible to know whether normal faulting at Osumi was accompanied or not by block tilting. As mentioned above, the southern Osumi normal faults likely abut against the Amami-Kagoshima Line between the Ata and Kikai calderas. Three fault zones can be distinguished. The northern fault zone separates the Osumi pluton from Palaeogene country rocks exposed to the south of the Kanoya plain, a flat-lying depression covered by Quaternary volcanic deposits of

302

O. FABBRICTAL.

Fig. 3. Simplified geological map of the southern part of the Osumi peninsula.

unknown thickness. At least two faulting stages are recognized: a N30°E-trending, NW-dipping fault of unknown kinematics (Oba 1961) is displaced by a N65°E-trending, north-dipping normal fault (Fig, 3). The central and southern

fault zones are well exposed and can be mapped across the entire length of the pluton. Along the central and southern fault zones, striations borne by fault planes indicate predominant dip-slip (Fabbri et al. 1997). Slip-sense indicators testify

Fig. 4. Interpretative cross-section across the Osumi pluton and associated fault system (no vertical exaggeration).

303

TRANSTENSION IN THE SW JAPAN ARC SYSTEM

to a normal slip component on all planes. The strike-slip component, which remains minor, is equally distributed between dextral and sinistral senses. In a few localities along the central fault zone, pseudotachylite veins are injected along or obliquely to the fault planes (Fabbri et al. 2000). The sense of shear indicated by the arrangement of these veins is always a normal one. During the course of the present study, 40 Ar/39Ar laser probe ages were determined both on biotite from the host granodiorite and on a pseudotachylite vein in the central fault zone (Table 1; for details on experimental procedure, see Monie et al. 1997). A single biotite grain from the granodiorite was progressively degassed using a defocused laser beam. It produces a partially discordant age spectrum (Fig. 5a) for which ages progressively increase from about 10 Ma until a plateau date of 14.1 + 0.4 Ma calculated for 58% of the argon released. In the isotope correlation plot (Fig. 5b), the data display a good linear trend with an intercept age of 14.5 ± 0.4 Ma and an initial 40Ar/36Ar ratio that is lower than the present-day air value and results from the partial resetting of the biotite evidenced by the first heating steps. It is noticeable that this new 40Ar/39Ar biotite age is fully consistent with a K-Ar age of 14.5 ± 0.3 Ma obtained on a nearby rhyolite (Fabbri, unpublished data; Table 1). It is interpreted to record cooling of the Osumi pluton through 300 °C. Pseudotachylite has been investigated in two ways. First, a small glass fragment optically free of any inclusions was progressively degassed. The corresponding age spectrum (Fig. 5c) shows apparent ages ranging from 10.0 ± 3.1 Ma to 18.6 ±3.7 Ma and a plateau age of 13.0 + 0.4 Ma for the less discordant portion of the spectrum. The isotope correlation plot gives

an intercept age of 13.2 ± 0.4 Ma with an atmospheric initial40Ar/36Ar ratio (Fig. 5d). In a second experiment, a series of in situ laser ablations was performed across a 1 cm wide pseudotachylite vein using a focused laser beam. In the host granite, biotite in contact with the vein gives ages scattering from 12.9 ± 0.4 Ma to 19.8 ± 0.8 Ma (Fig. 5e). In the vein, 11 analyses yield ages ranging from 13.1+0.3 Ma to 15.3 ± 0.4 Ma. This distribution is interpreted to result from the presence of numerous clasts (mainly quartz and feldspar) in the glassy matrix that has trapped a minor amount of excess argon during the formation of the vein. In the isotope correlation plot (Fig. 5f), the data are scattered along a line giving an intercept age of 13.2 ± 0.4 Ma and an initial 40Ar/3feAr ratio of 362 + 26 that is indicative of excess argon contamination. The presence of excess argon in pseudotachylite veins is a common feature (e.g. Kelley et al. 1994) due to the fact that their injection along cracks occurs nearly instantaneously, allowing isotopic exchanges only at a very small scale. Excess argon released from the vein has also been trapped by host rock biotite immediately in contact with the vein. More than 1 mm from this contact, biotite displays an age of 12.9 ± 0.4 Ma, which is consistent with the minimum age of the vein, with the intercept age in the isotope correlation plot and with the data reported above concerning the analysed pseudotachylite glass fragment. Therefore, we consider that these ages close to 13 Ma represent the best estimate for the pseudotachylite formation. Compared with published cooling ages obtained on the Osumi pluton (K-Ar on biotite, Shibata 1978; fission tracks on zircon and apatite, Miyachi 1985) and with the new 40Ar/39Ar biotite age on this pluton (this study), the 40Ar/39Ar dates of the

Table 1. Synthesis of ages obtained on pseudotachylite veins and host granodiorite at Osumi Method

Sample number

Rock type

Dated material

Age ± 2o- (Ma)

14.54 + 0.33 14.1 + 0.4 plateau age: 13.0 + 0.4 isochron: 13.2 ± 0.4 minimum ages: 12.9 + 0.4 & 13.1 + 0.3 isochron: 13.2 ± 0.4

K-Ar* Ar/39Ar 40 Ar/39Ar

OF-1 96080303 Q13

rhyolite granodiorite pseudotachylite vein

whole rock biotite glassy matrix

40

Q14

pseudotachylite vein

profile across the vein

40

Ar/39Ar (localized fusions)

*K-Ar whole-rock age of a rhyolite is an unpublished datum.

304

O. FABBRICTAZ,

Fig. 5. Results of 40Ar/39Ar laser probe dating of Osumi granodiorite and pseudotachylite. (a) Age spectrum of a single biotite grain from the host granodiorite. (b) Isotope correlation plot corresponding to (a), (c) Age spectrum of a small glassy, clast-free, pseudotachylite fragment from the central fault zone, (d) Isotope correlation plot corresponding to (c). (e) In situ laser ablations carried out with a focused laser beam across a 1 cm thick pseudotachylite vein. Partially discordant ages reflect the influence of clasts in the glassy matrix. Minimum values of 12.9-13.1 Ma are likely the less influenced ages, (f) Isotope correlation plot corresponding to (e).

pseudotachylites indicate that the deformation took place coevally with the cooling of the pluton (Fig. 6). According to the Research Group for Active Faults of Japan (1991) and our own observations, with the exception of a single isolated left-lateral strike-slip fault located in the northeastern part of the pluton

(Fig. 3), the faults at Osumi are currently inactive. Quaternary sediments of the Kanoya plain seal faults of the northern zone (Fig. 3). The precise age of such unfaulted sediments remains unknown but can be estimated at 2 Ma (Research Group for Active Tectonics in Kyushu 1989). If this age is confirmed, it

TRANSTENSION IN THE SW JAPAN ARC SYSTEM

Fig. 6. Cooling ages of the Osumi granodiorite versus formation ages of Osumi pseudotachylites. K-Ar ages on biotites from the granodiorite are from Shibata (1978). Zircon and apatite fission track ages are from Miyachi (1985).

means that the faults at Osumi have been inactive for the past two million years. Hitoyoshi-Ichifusa region Like the Osumi peninsula, the northeastern part of the Ichifusa area is occupied by a 14 Ma pluton intruded into Cretaceous to Oligocene sandstones and mudstones of the Shimanto Group (Fig. 7). The strata of the Shimanto Group are arranged in kilometre-scale thrust sheets bound by N60°E-trending reverse faults (Saito et al. 1996; for simplicity, the reverse faults are not represented on Fig. 7). Thrust planes dip weakly to moderately (0-55°) northwestwards. Emplacement of the Ichifusa pluton clearly post-dates reverse faulting and thrust sheet formation. Both the pluton and the Shimanto rocks are cut by numerous normal faults trending N60°E and dipping 40-75° northwestwards (Saito et al. 1996). The largest dips (70-75°) are observed for faults inside the pluton whereas the smallest ones (40-50°) are found for faults in the Shimanto Group. Age of normal faulting is post-14 Ma but cannot be determined more precisely. Normal faulting also affects the southwestern part of the Hitoyoshi-Ichifusa area. A conspicuous N50°E-trending, NW-dipping normal fault separates uplifted strata of the Shimanto Group at the footwall from Pliocene to Quaternary volcanic, lacustrine or fluviatile deposits of the Hitoyoshi basin at the hanging wall. To the

305

north or NW, the Plio-Quaternary deposits unconformably overlap the Shimanto Group. The Hitoyoshi basin thus appears as a halfgraben. Plio-Quaternary deposits are not accurately dated. The age of the oldest strata in the graben is estimated at 3 Ma (Kano et al. 1991). According to the Research Group for Active Faults of Japan (1991), faults bounding the graben have displaced Holocene terrace deposits and thus are currently active. Because of poor exposure quality, fault-slip data from the Hitoyoshi basin or from the Ichifusa pluton area are very scarce. At only two localities, Pliocene deposits show steeply dipping faults trending N45°E ± 20° (Tokushige & Fabbri 1996). Slip indicators testify to either normal slip (NW-SE extension) or oblique slip with a right-lateral component (NW-SE extension combined with NE-SW shortening). Beppu region The Beppu area is almost entirely covered by Pliocene to Quaternary lava flows, pyroclastic deposits and subordinate lacustrine or fluviatile deposits (Kamata 1989; Mizuno 1992). These strata are cut by a series of east-west normal faults defining a graben (Kamata 1989; Chida 1992). Outside the graben, pre-Miocene basement rocks are locally exposed, whereas, inside the graben, they have subsided to about 3000 m below the surface, as shown by Bouguer gravity anomaly, seismic reflection profiling and drilling. Based on seismic reflection profiles, Yusa et al. (1992), Yamakita et al. (1995) and Ito et al. (1996) interpreted the Beppu graben as a half-graben developed above the northwarddipping MTL reactivated as a master detachment fault during Plio-Quaternary times (Fig. 8). Timing is not accurately known, but the oldest rocks drilled in the Beppu graben are 6 Ma old, thus providing a maximum age for the inception of the subsidence in the half-graben. Geodetic measurements (Tada 1993; Nishimura et al. 1999), faulting of Holocene volcanic deposits and historical seismicity (Research Group for Active Tectonics in Kyushu 1989; Research Group for Active Faults of Japan 1991) show that extension is still active and is directed north-south. Summary Extension-related structures in the three studied areas show the following characteristics: 1 most normal faults, especially the laterally consistent ones, strike N50° to N90°E and

Fig. 7. Simplified geological map and cross-sections of the Hitoyoshi-Ichifusa area and associated fault system. Partly after Saito et al. (1996). For simplicity, traces of thrust faults are omitted. No vertical exaggeration on cross-sections.

TRANSTENSION IN THE SW JAPAN ARC SYSTEM

307

Fig. 8. Section across the Beppu Bay (section C-C' on Fig. 1). Modified after Yamakita et al (1995). SL, sea level.

dip moderately to steeply towards the north or the NW; 2 in the two areas of Hitoyoshi-Ichifusa and Beppu, a half-graben developed upon a master northward-dipping normal fault, suggesting block tilting; 3 activity of normal faults has a maximum age of 14 Ma, and took place at 13 Ma at Osumi; 4 normal faults are still active at Beppu and at Hitoyoshi, but have been inactive since about 2 Ma at Osumi. Proposed model of evolution of the Okinawa-Kyushu junction area Transtension in the northern Okinawa trough—Kyushu region At the junction between the Okinawa trough back-arc basin and the SW Japan arc, the extension-related structures are characterized in crosssection by an asymmetric half-graben structure and, in map view, by a left-stepping en echelon arrangement. This distinctive geometry requires a combination of extension and right-lateral shear along a N10° to N30°E direction (Fig. 9). The right-lateral shear implies that the oceanward side of the arc has been moving south to SW relative to the back-arc domain (continent). Such a right-lateral motion has already been proposed in the northern and central parts of the Okinawa trough by Fournier et al. (2001) to account for the present-day kinematics and more precisely to explain the right-lateral motion along north-south- to N20°E-trending nodal planes of focal mechanisms for earthquakes in the region. This general motion also accounts for Quaternary right-lateral motion along the MTL in western Shikoku and eastern Kyushu, and along N20° to N30°E-trending faults in central Kyushu (SW of Aso caldera, Fig. 1). The scenario proposed by Fournier et al.

(2001) associates a perpendicular opening in the southern part of the Okinawa trough and an oblique transtensional opening in the northern part of the trough (Fig. 9). Accommodation of extension through reactivation of thrust faults The spatial association between Miocene or postMiocene extensional structures and pre-Miocene shortening structures is obvious in the Beppu Bay and Hitoyoshi-Ichifusa areas. As already noted by Yusa et al. (1992), Yamakita et al. (1995) and Ito et al. (1996), the Beppu graben is in the hanging wall of the MTL, which was reactivated as a normal fault in Plio-Quaternary times. Reverse motion along the MTL near Beppu dates back to the Oligocene or earlier (Yamakita et al. 1995). Normal faults also developed in the vicinity of the NTL (Fig. 1), a major fault whose reverse motion accumulated before the Middle Miocene (as attested by the cross-cutting Ichifusa pluton). Though still to be ascertained by field or subsurface evidence, reactivation of the NTL as a detachment fault could explain the development of normal faults in the pre-Miocene Shimanto Group as well as in the Ichifusa pluton. In the Osumi area, the spatial link between Miocene or younger normal faults and the pre-Miocene thrust faults is less obvious than in the two other areas because of the widespread Quaternary cover of the Kanoya plain and of the sea. Geophysical investigations are obviously needed to allow a better understanding of the possible spatial or genetic links between the two kinds of faults in the Osumi area. Cross-section model Figure 10 depicts a schematic cross-section of the present-day convergence between the Philippine

308

O. FABBRICTAL.

Fig. 9. Perpendicular extension in the southern part of the Okinawa trough versus oblique opening in the northern part of the trough and in Kyushu. Modified after Sibuet et al (1987) and Founder et al (2001). The arrow represents the relative direction of convergence between the Philippine Sea and Eurasia plates. Absolute ages are the estimates for the periods of extension in the three areas of Osumi (13-2 Ma), Hitoyoshi-Ichifusa (13 Ma to present) and Beppu (6 Ma to Present). Inset: Geometric relationship between the direction of extension (large divergent arrows) and the direction of relative displacement of the Ryukyu arc relative to Eurasia. In the southern Okinawa trough, the direction of extension is parallel to that of relative displacement, whereas it trends halfway between the direction of relative displacement and the perpendicular to the rift, as shown by Withjack & Jamison (1986) and Tron & Brun (1991).

Sea and Eurasia plates. We propose that the normal faults in southern Kyushu (Osumi on the section) and those bounding the halfgrabens in the northern Okinawa trough extend downwards through the entire brittle upper crust and merge into a partial attachment or coupling zone at the transition between the ductile lower crust and the upper brittle crust (Tikoff et al 2002). The attachment zone could explain present-day and past extension in

apparently scattered individual basins in the junction area, in a way similar to that described in eastern Greenland by Larsen (1988), with the difference that, in the present case, a strike-slip component is also recorded. Furthermore, it could provide kinematic linkage between the upper crust undergoing heterogeneously distributed brittle deformation and the lower crust characterized by distributed ductile deformation. Toward the continent, the attachment zone could

Fig. 10. Idealized cross-section showing partial attachment/coupling (in the sense of Tikoff et al 2002) at the transition zone between the brittle upper crust and ductile lower crust of the Eurasia lithosphere (section D-D' on Fig. 1; no vertical exaggeration). The crustal thickness of the arc (26-30 km) is after Iwasaki et al. (1990). The dip of the Wadati-Benioff plane is after Nagamune & Tashiro (1989). Yangsan fault (YF), Tsushima fault (TF) and Nakadori fault (NF) are regional strike-slip faults.

310

O. FABBRI £7 AL.

either cross the Eurasia lithosphere, in accordance with the lithosphere-scale simple shear model (Wernicke 1985), or merge with a lithosphere-scale vertical strike-slip zone, following the model of Teyssier & Tikoff (1998). In this last model, the lithospheric strike-slip zone could lie beneath the noth-south to NE-SW strike-slip faults known in the vicinity of the Goto islands and further north (Figs 1 & 10). Alkaline basalts exposed in NW Kyushu and in the Goto islands and radiometrically dated between 12 and 1 Ma (Nakamura et al. 1985; Kano et al. 1991), which could reflect the upwelling of the asthenospheric mantle beneath the Goto sub-basin where the extension is active, would favour the lithosphere-scale simple shear model. Additional data, especially geophysical, are needed to help make precise the geometry of the attachment zone toward the continent. Conclusion Mapping of normal faults in the KyushuOkinawa trough junction area, along with the dating of the fault motion, leads to the following results. Firstly, it shows that extension started in the junction area as early as Middle Miocene times (c. 13 Ma), as already suggested off-shore by Letouzey & Kimura (1985, 1986). Secondly, it confirms that extension was combined with dextral strike-slip shear. Thirdly, it demonstrates that crustal stretching is not restricted to a narrow graben, but encompasses a wide area inside which localization of extension has evolved through time. The distributed transtensional deformation at the junction between a nascent back-arc rift and an arc, which appears to be a characteristic of the study area, was likely favoured by the normal-slip reactivation of low-angle thrust faults, which are widespread in the fore-arc and arc domains. References ACOCELLA, V., FACCENNA, C., FUNICIELLO, R. & ROSSETTI, F. 1999. Sand-box modelling of basement-controlled transfer zones in extensional domains. Terra Nova, 11, 149—156. AIBA, J. & SEKIYA, E. 1979. Distribution and characteristics of the Neogene sedimentary basins around the Nansei-Shoto (Ryukyu Islands). Journal of the Japanese Association of Petroleum Technologists, 44, 229-340. ANGELIER, J. & HUCHON, P. 1987. Tectonic records of convergence changes in a collision area: the Boso and Miura peninsulas, central Japan. Earth and Planetary Science Letters, 81, 397-408. ANGELIER, J., COLLETTA, B., CHOROWICZ, J., ORTLIEB, L. & RANGIN, C. 1981. Fault tectonics

of the Baja California Peninsula and the opening of the Sea of Cortez, Mexico. Journal of Structural Geology, 3, 347-357. ANGELIER, J., BARRIER, E. & CHU, H.T. 1986. Plate collision and palaeostress trajectories in a foldthrust belt: the foothills of Taiwan. Tectonophysics, 125, 161-178. BASILE, C. & BRUN, J.P. 1999. Transtensional faulting patterns ranging from pull-apart basins to transform continental margins: an experimental investigation. Journal of Structural Geology, 21, 23-38. BENES, V. & SCOTT, S.D. 1996. Oblique rifting in the Havre Trough and its propagation into the continental margin of New Zealand: comparison with analogue experiments. Marine Geophysical Research, 18, 189-201. BOCCALETTI, M., CELLO, G. & TORTORICI, L. 1987. Transtensional tectonics in the Sicily Channel. Journal of Structural Geology, 9, 869-876. CARESS, D.W. 1991. Structural trends and backarc extension in the Havre Trough. Geophysical Research Letters, 18, 853-856. CHIDA, N. 1992. Active faults in Central Kyushu, Southwest Japan - Quaternary faulting along the Median Tectonic Line in Kyushu. Memoirs of the Geological Society of Japan, 40, 39-51. DAUTEUIL, O. & BRUN, J.P. 1993. Oblique rifting in a low-spreading ridge. Nature, 361, 145-148. DAUTEUIL, O. & BRUN, J.P. 1996. Deformation partitioning in a slow spreading ridge undergoing oblique extension: Mohns Ridge, Norwegian Sea. Tectonics, 15, 870-884. DELTEIL, J., RUELLAN, E., WRIGHT, I. & MATSUMOTO, T. 2002. Structure and structural development of the Havre Trough (SW Pacific). Journal of Geophysical Research, 107, (10.1029/2001JB000494). DeMETS, C., GORDON, R.G., ARGUS, D.F. & STEIN, S. 1990. Current plate motions. Geophysical Journal International, 101, 425-478. EGUCHI, T. & UYEDA, S. 1983. Seismotectonics of the Okinawa trough and Ryukyu arc. Memoirs of the Geological Society of China, 5, 189-210. EL-FIKY, G.S., KATO, T. & OWARE, E.N. 1999. Crustal deformation and interplate coupling in the Shikoku district, Japan, as seen from continuous GPS observation. Tectonophysics, 314, 387-399. FABBRI, O. & FOURNIER, M. 1999. Extension in the southern Ryukyu arc (Japan). Link with oblique subduction and back-arc rifting. Tectonics, 18, 486-497. FABBRI, O., TOKUSHIGE, H. & HAYAMIZU, M. 1997. Normal faulting in the Middle Miocene Osumi granodioritic pluton, southern Kyushu, and its significance. Journal of the Geological Society of Japan, 103, 141-151. FABBRI, O., LIN, A.M. & TOKUSHIGE, H. 2000. Coeval formation of cataclasite and pseudotachylyte in a Miocene forearc granodiorite, southern Kyushu, Japan. Journal of Structural Geology, 22, 10151025. FOURNIER, M., FABBRI, O., ANGELIER, J. & CADET, J.P. 2001. Regional seismicity and onland deformation in the Ryukyu arc: implications on the kinematics

TRANSTENSION IN THE SW JAPAN ARC SYSTEM of opening of the Okinawa Trough. Journal of Geophysical Research, 106, 13751-13768. HIRATA, N., KINOSHITA, H. et al 1991. Report on DELP 1988 cruises in the Okinawa Trough, Part III: Crustal structure of the southern Okinawa Trough. Bulletin of the Earthquake Research Institute, 66, 37-70. ILLIES, J.H. 1977. Ancient and recent rifting in the Rhine Graben. Geologic en Mijnbouw, 56, 329-350. IMANISHI, M., KIM ATA, F. et al. 1996. Horizontal displacements by GPS measurements at the OkinawaSakishima Islands (1994-1995). Journal of the Seismological Society of Japan, 49, 417—421. ITO, T., IKAWA, T., YAMAKITA, S. & MAEDA, T. 1996. Gently dipping Median Tectonic Line (MTL) revealed by recent seismic reflection studies, southwest Japan. Tectonophysics, 264, 51-63. IWASAKI, T., HIRATA, N. et al. 1990. Crustal and upper mantle structure in the Ryukyu Island Arc deduced from deep seismic sounding. Geophysical Journal International, 102, 631-651. KAMATA, H. 1989. Volcanic and structural history of the Hohi volcanic zone, central Kyushu, Japan. Bulletin de Volcanologie, 51, 315-332. KAMATA, H. 1998. Quaternary volcanic front at the junction of the Southwest Japan arc and the Ryukyu arc. Journal of Asian Earth Sciences, 16, 67-75. KANO, K., KATO, H., YANAGISAWA, Y. & YOSHIDA, F. 1991. Stratigraphy and Geologic History of the Cenozoic of Japan. Reports of the Geological Survey of Japan, 274. KAO, H. & CHEN, W.P. 1991. Earthquakes along the Kyushu-Ryukyu arc: Strain segmentation, lateral compression, and the thermomechanical state of the plate interface. Journal of Geophysical Research, 96, 21443-21485. KELLEY, S.P., REDDY, S.M. & MADDOCK, R. 1994. Laser-probe 40Ar/39Ar investigation of a pseudotachylyte and its host rock from the Outer Isles thrust, Scotland. Geology, 22, 443-446. KlMURA, K., IWAYA, T., MlMURA, K., SATO, Y., SATO,

T., SUZUKI, Y. & SAKAMAKI, Y. 1991. Geological Sheet Map of the Osuzuyama District (1/50000) and Explanatory Notice. Geological Survey of Japan. KIMURA, M. 1985. Back-arc rifting in the Okinawa Trough. Marine and Petroleum Geology, 2, 222-240. LARSEN, P.H. 1988. Relay structures in a Lower Permian basement-involved extension system, East greenland. Journal of Structural Geology, 10, 3-8. LEE, C.S., SHOR, G.G., BIBEE, L.D., Lu, R.S. & HILDE, T. 1980. Okinawa Trough: genesis of a back-arc basin. Marine Geology, 35, 219-241. LETOUZEY, J. & KIMURA, M. 1985. Okinawa Trough genesis: structure and evolution of a backarc basin developed in a continent. Marine and Petroleum Geology, 2, 111-130. LETOUZEY, J. & KIMURA, M. 1986. The Okinawa Trough: genesis of a back-arc basin developing along a continental margin. Tectonophysics, 125, 209-230.

311

MART, Y. & DAUTEUIL, O. 2000. Analogue experiments of propagation of oblique rifts. Tectonophysics, 316, 121-132. MATSUDA, T. 1980. Traces of Izu Peninsula - its direction of movement during the recent several million years. The Earth Monthly, 2, 164-168. MIYACHI, M. 1985. Fission track ages of some granitic rocks in the Outer Zone of Kyushu, Japan. Journal of the Japanese Association of Mineralogists, Petrologists and Economic Geologists, 80, 406-409. MIZUNO, K. 1992. Age and tectonic development of the Plio-Pleistocene sedimentary basins along the Median Tectonic Line, Southwest Japan. Memoirs of the Geological Society of Japan, 40, 1-14. MONIE, P., CABY, R. & ARTHAUD, M.H. 1997. The Neoproterozoic Brasiliano orogeny in northeast Brazil: 40Ar/39Ar and petrostructural data from Ceara. Precambrian Research, 81, 241-264. MURATA, A. 1981. Large decke structures in the Kurosegawa and Sambosan terrains in Kyushu, southwest Japan. Journal of the Faculty of Science, University of Tokyo, 20, 277-293. MURATA, A. 1987. Conical folds in the Hitoyoshi bending, south Kyushu, formed by the clockwise rotation of the southwest Japan arc. Journal of the Geological Society of Japan, 93, 91 — 105. MURATA, A. 1991. Duplex structures in the Uchinohae Formation in the Shimanto Terrane, Kyushu, Southwest Japan. Journal of the Geological Society of Japan, 97, 39-52. NAGAMUNE, T. & TASHIRO, H. 1989. Shape of the Wadati-Benioff zone beneath Kyushu, Japan. Journal of the Seismological Society of Japan, 42, 13-19. NAKADA, S. & KAMATA, H. 1991. Temporal change in chemistry of magma source under Central Kyushu, Southwest Japan: progressive contamination of mantle wedge. Bulletin of Volcanology, 53, 182-194. NAKAMURA, E., CAMPBELL, I.H. & SUN, S. 1985. The influence of subduction processes on the geochemistry of Japanese basalts. Nature, 316, 55-58. NASH, D.F. 1979. The geological development of the North Okinawa Trough area from Neogene times to recent. Journal of the Japanese Association of Petroleum Technologists, 44, 109—119. NISHI, H. 1988. Structural analysis of part of the Shimanto accretionary complex, Kyushu, Japan, based on planktonic foraminiferal zonation. Modern Geology, 12, 47-69. NISHIMURA, S., ANDO, M. & MIYAZAKI, S. 1999. Interplate coupling along the Nankai Trough and southeastward motion along southern part of Kyushu. Journal of the Seismological Society of Japan, 51, 443-456. NOTSU, K., ONO, K. & SOYA, T. 1987. Strontium isotopic relations of bimodal volcanic rocks at Kikai volcano in the Ryukyu arc, Japan. Geology, 15, 345-348. NOZAWA, T. & OTA, R. 1967. Geological Sheet Map of the Uchinoura District (1/50 000) and Explanatory Notice. Geological Survey of Japan. OB A, N. 1961. Petrological study of Osumi granite, especially its heterogeneity and schistose structure.

312

O. FABBRI£7AL.

Miscellaneous Research Reports, Institute of Natural Resources, 54-55, 191-201. OB A, N. 1977. Emplacement of granitic rocks in the Outer Zone of Southwest Japan and geological significance. Journal of Geology, 85, 383-393. OB ATA, H. 1961. Morphography of Osumi peninsula. Miscellaneous Research Reports, Institute of Natural Resources, 54, 151-160. OGAWAUCHI, Y. & IWAMATSU, A. 1986. Geological structures of the Shimanto Terrain in the Osumi Peninsula, Southern Kyushu. Memoirs of the Faculty of Science, Kagoshima University, 19, 149-169. PARK, J.O., TOKUYAMA, H., SHINOHARA, M., SUYEHIRO, K. & TAIRA, A. 1998. Seismic record of tectonic evolution and back-arc rifting in the southern Ryukyu island arc system. Tectonophysics, 294,21-42. PARSON, L.M. & WRIGHT, I.C. 1996. The Lau-HavreTaupo backarc basin: a southward-propagating, multistage evolution from rifting to spreading. Tectonophysics, 263, 1-22. RESEARCH GROUP FOR ACTIVE FAULTS OF JAPAN 1991. Active Faults in Japan. Sheet maps and inventories. Revised Edition. Tokyo University Press, Tokyo. RESEARCH GROUP FOR ACTIVE TECTONICS IN KYUSHU 1989. Active Tectonics of Kyushu. Tokyo University Press, Tokyo. SAITO, M., KIMURA, K., NAITO, K. & SAKAI, A. 1996. Geological Sheet Map of the Shiibamura District (1/50000) and Explanatory Notice. Geological Survey of Japan. SAKAI, T. & KANMERA, K. 1981. Stratigraphy of the Shimanto terrane and the tectono-stratigraphic setting of greenstones of the northern part of Miyazaki Prefecture, Kyushu. Science Reports, Faculty of Science, Kyushu University, 14, 31-48. SCHREURS, G. & COLLETTA, B. 1998. Analog modelling of faulting in zones of continental transpression and transtension. In: HOLDSWORTH, R.E., STRACHAN, R.A. & DEWEY, J.F. (eds) Continental Transpressional and Transtensional Tectonics. Geological Society, London, Special Publications, 135, 5-26. SCHUMACHER, M.E. 2002. Upper Rhine graben: role of preexisting structures during rift evolution. Tectonics, 21, 1 (10.1029/2001TC900022). SENO, T. & MARUYAMA, S. 1984. Paleogeographic reconstruction and origin of the Philippine Sea. Tectonophysics, 102, 53-84. SENO, T., STEIN, S. & GRIPP, A.E. 1993. A model for the motion of the Philippine Sea plate consistent with NUVEL-1 and geological data. Journal of Geophysical Research, 98, 17941-17948. SHIBATA, K. 1978. Contemporaneity of Tertiary granites in the Outer Zone of southwest Japan. Bulletin of the Geological Survey of Japan, 29, 551-554. SIBUET, J.C., LETOUZEY, J. et al 1987. Back-arc extension in the Okinawa Trough. Journal of Geophysical Research, 92, 14041-14063. SIBUET, J.C., Hsu, S.K., SHYU, C.T. & Liu, C.G. 1995. Structural and kinematic evolutions of the Okinawa Trough backarc basin. In: TAYLOR, B. (ed.)

Backarc Basins: Tectonics and Magmatism. Plenum Press, New York, 343-379. SIBUET, J.C., DEFFONTAINES, B. et al. 1998. Okinawa trough backarc basin: early tectonic and magmatic evolution. Journal of Geophysical Research, 103, 30245-30267. TAD A, T. 1993. Crustal deformation in central Kyushu, Japan, and its tectonic implication - rifting and spreading of the Beppu-Shimabara graben. Memoirs of the Geological Society of Japan, 41, 1-12. TAIRA, A., OKADA, H., WHITAKER, J.H.McD. & SMITH, A.J. 1982. The Shimanto Belt of Japan: Cretaceous-Lower Miocene active margin sedimentation. In: LEGGETT, J.K. (ed.) TrenchForearc Geology. Geological Society, London, Special Publications, 10, 5-26. TEYSSIER, C. & TIKOFF, B. 1998. Strike-slip partitioned transpression of the San Andreas fault system: a lithospheric scale approach. In: HOLDSWORTH, R.E., STRACHAN, R.A. & DEWEY, J.F. (eds) Continental Transpressional and Transtensional Tectonics. Geological Society, London, Special Publications, 135, 143-158. TIKOFF, B., TEYSSIER, C. & WATERS, C. 2002. Clutch tectonics and the partial attachment of lithospheric layers. In: BERTOTTI, G., SCHULMANN, K. & CLOETINGH, S. (eds) Continental Collision and the Tectono-Sedimentary Evolution of Forelands. European Geophysical Society Special Publications, 1, 119-144. TOKUSHIGE, H. & FABBRI, O. 1996. Mesofaults and associated stress field in the Late Miocene to Pliocene forearc deposits of the Miyazaki district, southeast Kyushu, Japan. Journal of the Geological Society of Japan, 102, 622-634. TRON, V. & BRUN, J.P. 1991. Experiments on oblique rifting in brittle-ductile systems. Tectonophysics, 188,71-84. WERNICKE, B. 1985. Uniform-sense normal simple shear of the continental lithosphere. Canadian Journal of Earth Sciences, 22, 108-125. WITHJACK, M. & JAMISON, W. 1986. Deformation produced by oblique rifting. Tectonophysics, 126, 99-124. WRIGHT, I.C. 1993. Pre-spreading rifting and heterogeneous volcanism in the southern Havre Trough backarc basin. Marine Geology, 113, 179-200. WRIGHT, I.e., PARSON, L.M. & GAMBLE, J.A. 1996. Evolution and interaction of migrating cross-arc volcanism and backarc rifting: an example from the southern Havre Trough (35°20-37°S). Journal of Geophysical Research, 101, 2207122086. YAMAKITA, S., ITO, T., TANAKA, H. & WATANABE, H. 1995. Early Oligocene top-to-the-west motion along the Sashu fault, a low-angle oblique thrust of the paleo-Median Tectonic Line, east Kyushu, Japan. Journal of the Geological Society of Japan, 101, 978-988. YUSA, Y., TAKEMURA, K., et al. 1992. Subsurface structure of Beppu Bay (Kyushu, Japan). Journal of the Seismological Society of Japan, 45, 199-212.

Isothermal decompression, partial melting and exhumation of deep continental crust DONNA L. WHITNEY, CHRISTIAN TEYSSIER & ANNIA K. FA YON Department of Geology and Geophysics, University of Minnesota, Minneapolis, Minnesota 55455, USA (e-mail: [email protected]) Abstract: Decompression of deep, hot continental crust is the primary mechanism of crustal melting, with major consequences for the geodynamics of orogens. Decompression within thickened continental crust may be initiated by processes driven from above (erosion, tectonic denudation) and/or below (crust/lithosphere thinning, buoyant rise of deep crust). On a larger scale, decompression of subducted continental crust may add material, including melt, to the overlying, non-subducting plate. This mechanism has the potential to produce large amounts of melt because fertile material is continually conveyed into the mantle, where it eventually buoyantly ascends and melts. Decompression-driven melting of continental crust may account for the high melt fractions (>20 vol.%) and great thickness (20-30 km) inferred for the partially molten layer in erogenic crust. When high melt volumes are present in the crust and/or the thickness of the partially molten layer is large, the subsequent thermo-mechanical evolution of orogens is strongly influenced by lateral (channel) and vertical (buoyant) crustal flow. For both lateral and vertical flow, the presence of melt decouples deep crust from upper crust, and continental crust from mantle lithosphere. A major consequence of vertical crustal flow is the generation of migmatite-cored gneiss domes that riddle most orogens. High-grade rocks in many domes record pressure-temperature-time (P-T-t) paths indicating near-isothermal decompression followed by cooling from T> 700 °C to T < 350 °C in <2-5 Ma. Diapiric ascent of partially molten crust accounts for the decompression rate and magnitude required to maintain a near-isothermal path. We propose that gneiss domes are a signature of decompression and crustal melting, and are therefore fundamental structures for understanding the thermo-mechanical evolution of continental crust during orogeny.

Construction of an orogen involves crustal thickening, which occurs by tectonic shortening, addition of mantle-derived magma and/or thermal uplift of the continental lithosphere of the non-subducting plate as it interacts with the subducting/colliding plate. Crustal thickening may also occur by transfer of deeply subducted continental material from the downgoing plate to the lower crust of the non-subducting plate, During and following thickening, the nature and magnitude of thermo-mechanical links between mantle lithosphere and continental crust and between deep crust and upper crust depend in large part on the extent of partial melting of the deep crust. It is therefore important to document the location (depth), mechanisms and magnitude of crustal melting to understand the degree of coupling or decoupling between these different lithospheric layers. Volumetrically large amounts of melt are/ were present in modern and ancient orogens. Seismic and magnetotelluric data have been

interpreted to indicate the presence of melt in the deep crust of young orogens (Pyrenees: Pous et al. 1995; Tibet: Chen et al. 1996; Nelson et al. 1996; Him et al 1997; Central Andes and Tibet: Schilling & Partzsch 2001). Geophysical data are consistent with ~20 vol.% melt, which is likely crustally derived (Schilling & Partzsch 2001), and which may coalesce to form leucogranites (Brown et al. 1996). In this chapter, we explore the idea that some of the partially molten crust in collisional orogens may be derived from subducted continental material. The decompression of deeply buried or subducted continental crust is a possible mechanism for large-magnitude, near-isothermal decompression and extensive partial melting, The following are some important questions related to the origin and geodynamic evolution of partially molten crust: What mechanism(s) drive large-scale crustal melting? What are the thermo-mechanical links between continental crust in the subducting v. non-subducting plates

From: GROCOTT, J., MCCAFFREY, K. J. W., TAYLOR, G. & TIKOFF B. (eds) 2004. Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 313-326. 0305-8719/047$ 15 © The Geological Society of London 2004.

314

D. L.WHITNEY ET AL.

of a collision zone? What is the thermal, structural and petrological fate of partially molten crust? Partial melting of continental crust may occur by heating and/or decompression. In this chapter, we consider the evidence for and effectiveness of different mechanisms for generating melt in the deep crust. We focus on decompression because exhumed migmatite complexes - in particular, those that experienced mica dehydration melting - are commonly characterized by hightemperature, near-isothermal decompression paths and show evidence for progressive melting from high pressure (garnet stability field) to lower pressures (cordierite stability field). In addition, decompression at elevated temperatures (>700 °C) can cause dehydration melting of common crustal protoliths (e.g. mica-rich gneiss) to produce high melt fractions (e.g. Brown 1994). The results of thermochronology and thermal modelling can be used to evaluate the conditions necessary to maintain near-isothermal conditions during decompression and to examine the thermal effects of partially molten regions that have risen from depth (>30km) to shallow (<15 km) levels of the crust, as in the case of many migmatite-cored gneiss domes. We discuss in particular the possibility that regions of melt-bearing crust may rise buoyantly as migmatitic diapirs to form gneiss domes: domal structures cored by migmatite, orthogneiss and granitoid that occur in most orogens.

Crustal melting A viable mechanism for melting in the deep crust must account for: (1) the amount of anatectic leucosome (>20 vol.%) observed in many exhumed mid- to lower crustal terranes made up of migmatites and other crustally derived magmatic bodies (e.g. Brown 1994; Nyman et al 1995) as well as inferred for active orogens (e.g. Schilling & Partzsch 2001); (2) the near-isothermal decompression path recorded by migmatites and other high-grade metamorphic rocks; and (3) the observation that, in the case of many upper amphibolite facies migmatite terranes, the melt has not segregated on a large scale from its source rocks. Because high-temperature decompression paths are documented for many regional-scale migmatite terranes, this chapter focuses on melting during decompression: the conditions and mechanisms of decompression, and the consequences of this process.

Melt generation In an active orogen, partial melting of crustal rocks may occur at various stages during prograde heating by water-saturated and waterundersaturated (dehydration) melting, and/or during isothermal decompression by dehydration melting. Although water-saturated melting is commonly discounted because of the predicted low porosity (<1%) of deep crustal rocks, water-saturated melting may account for partial melting of mid-crustal terranes that are infiltrated by water-rich fluid from crystallizing plutons (Brown 1979; Montel et al. 1992) or deep-circulating meteoric water (e.g. Wickham 1987), particularly in cases where tonalitic or trondhjemitic leucosomes are produced from mica-bearing source rocks (Whitney & Irving 1994). In the absence of a major source of free fluid, watersaturated melting driven by release of aqueous fluids from dehydration of wet protoliths may produce low fractions (<10vol.%) of silicic melts, assuming ~ 1 wt% free water and the solubility of water in crustal melts (Holloway & Blank 1994). These melts will likely crystallize soon after segregation if temperature or pressure decrease, as predicted by the negative slope of water-saturated solidi in pressure-temperature (P-T) space. Vapour-absent (dehydration) melting can, however, produce greater amounts of granitic melt (20-60 vol.%; Clemens & Vielzeuf 1987; Vielzeuf & Holloway 1988) that may remain largely uncrystallized during ascent of the melt or decompression of the melt and host rock. Within a region, the melt fraction produced during dehydration melting is a function of the bulk composition of rocks that reached P-T conditions sufficient for melting, and the volume of fertile rock. Reaction curves for biotite dehydration melting have a steep slope, and so are encountered by P-T trajectories during steep decompression or during heating and initial decompression (Fig. 1). Once the solidus is crossed, paths tend to be subparallel to the solidus, on the melt-present side, perhaps owing to a combination of thermal buffering by the melt and dynamic effects of melt-enhanced decompression (discussed in later sections). The melt will not crystallize until it reaches much lower pressures because of the steep dP/dT slope of the solidus. Melting may also occur during the late stages of decompression as water is released from crystallizing melts, creating conditions for water-saturated melting (Thompson 2000). Melt segregation: mechanisms and scales The observation that migmatite terranes contain large fractions (20-40 vol.%) of crystallized

DECOMPRESSION MELTING AND EXHUMATION

315

shear zones (Brown & Solar 1998), or channels (Sawyer 2001). Most of these segregation/transport models involve consideration of how melt separates from its source and disperses. In the discussion below, we consider the implications for crustal dynamics of melt not leaving the system: that is, melt is segregated only on a local scale (1 cm to 100 m) and remains largely within the source rocks. Although some melt may escape, enough is retained for the migmatite complex to be buoyant. In this case, the most significant motion is not of the melt itself relative to its source (matrix), but of melt and matrix relative to less molten/more dense rocks. The generation and partial retention of melt may strongly influence the subsequent P-Tpath and dynamics of the partially molten rocks. We next consider mechanisms that could produce decompression melting of continental crust, before further discussing the implications for the thermo-mechanical evolution of orogens. Fig. 1. Pressure-temperature-time paths showing near-isothermal decompression of: (1) Thor-Odin gneiss dome, Shuswap complex, British Columbia (Norlander et al 2002); and (2) South Brittany migmatite terrane (Audren & Triboulet 1993). Various reference solidi are shown: water-saturated melting of a metapelitic rock (dashed curve); dehydration melting of muscovite gneiss (end-member muscovite + albite, and muscovite + intermediate plagioclase); and dehydration melting of biotite gneiss. Although shown as a line in the figure, biotite dehydration melting defines a region in P-T space, owing to solid solution in biotite. Most migmatite terrains and gneiss domes did not exceed the upper stability of biotite -I- plagioclase + quartz, as indicated by the lack of orthopyroxene. Because some migmatites/gneiss domes show evidence for late, low-pressure, high-temperature growth of cordierite, two equilibria involving cordierite are also shown (one solidus, one subsolidus).

Near-isothermal decompression paths and mechanisms

In thickened orogenic crust, the movement of high-grade metamorphic rocks from the mid- to lower crust towards the Earth's surface commonly occurs at near-isothermal conditions for a few to >10kbar of decompression (>33km) (Fig. 1). In some terranes, decompression may be preceded or accompanied by heating. High-temperature decompression paths have been demonstrated for migmatite terranes (e.g. Jones & Brown 1990; Whitney 1992; Audren & Triboulet 1993), including migmatite-cored gneiss domes (Norlander et al. 2002), and other high-grade rocks, including some ultrahigh-pressure terranes (e.g. Su-Lu, China: Wang et al 1993; Western Gneiss Region, Norway: Dunn & Medaris 1989). P-T paths estimated for many exhumed high-grade silicic melt that has segregated on a centimetre metamorphic rocks as well as those calculated scale but has not drained far from its source has by forward modelling for the thermal evolution fuelled a long debate about the connection, or of thickened crust (England & Thompson 1984) lack thereof, between anatectic migmatites and remain within ^50-80 °C of rmax during a subgranitoids (Brown 1994, and references stantial portion of the decompression path. therein). Much attention has been given to deter- These temperatures are within the uncertainty mining small-scale melt-segregation mechan- range of most metamorphic temperature calcuisms and discussing how silicic melt, once lations (e.g. geothermometry based on cation segregated from its source, might flow, coalesce exchange), so it is not possible to discern details and pool to form larger-scale magma bodies, of the paths other than that they remain at high such as orogenic leucogranites (Le Fort et al. temperature during decompression. 1987; Inger & Harris 1993). Possible mechanisms For average values of thermal conductivity, contributing to segregation and transport include mantle heat flux, crustal heat production and other compaction (McKenzie 1984) and flow through factors related to deformation and metamordykes/fractures (Clemens & Mawer 1992), phic reactions, isothermal decompression is not

316

D. L.WHITNEY ET AL

predicted by thermal models for slowly eroding (<1 mm a"1) orogens unless there is a gap in time (~10-30 Ma) between maximum crustal thickening and unroofing. This gap permits the generation of radiogenic heat from buried crust and relaxation of isotherms. For orogens with little or no gap in time between thickening and denudation/decompression, elevated temperatures can be maintained during decreasing pressure if decompression is rapid (^>1 mm a"1). In the following sections, we consider different decompression mechanisms (Fig. 2) and discuss whether each is likely to be a significant factor in maintaining elevated temperatures during decompression. We focus on mechanisms that account for decompression on a regional scale with magnitudes of > 10 km (and, in some cases, tens of kilometres), such as commonly observed in migmatite complexes, gneiss domes and ultrahigh-pressure metamorphic terranes, so that we can better understand thermomechanical links and feedback relationships among orogenic processes operating at the surface, at different levels within the continental crust and within the mantle. Erosion Unroofing via erosion may account for largemagnitude isothermal decompression if the erosional products are removed from the system or if there is a large lag time (>10Ma) between crustal thickening (deep burial) and the initiation of denudation. In the latter case, high-temperature decompression and partial melting may occur in part because of the added crustal heat production from buried radiogenic crust, following crustal thickening (England & Thompson 1986). Rapid erosion (>l-5 mm a"1) in concert with tectonism (uplift, faulting; Fig. 2a) may account for localized decompression, such as beneath alpine glaciers and deeply incised rivers (e.g. Zeitler et al 2001), although, in the latter case, many questions remain about how rivers or river systems evolve (move) through time and influence exhumation over a region much greater than that of the main incised channel. For regionalscale, large-magnitude decompression of deep crust, it is likely that other mechanisms are responsible for near-isothermal decompression, particularly in orogens in which there was no lag time between crustal thickening and thinning/ decompression. Low-angle normal faults Detachment systems can lead to exhumation/ decompression of deep-seated rocks by crustal

thinning during symmetric or asymmetric extension (Lister & Davis 1989). For example, asymmetric detachment systems may create a locus of lower crust upwelling that is offset relative to the normal fault break-away zone. Melting during this initial decompression is likely; however, given the low angle of the detachment, rocks cool as they are exhumed in the footwall of the detachment system, and therefore do not experience isothermal decompression (Fayon et al. 2002). Therefore, detachment faults are unlikely to account for the magnitude of decompression observed in gneiss domes and other high-pressure terranes unless the faults were originally higher angle (Buck 1988) or the rate of motion on detachments is unusually fast. Crustal thinning/collapse Regional crustal thinning may drive decompression of the deep crust during collapse of thickened crust (Rey 1993) (Fig. 2b), especially if thinning is localized in a narrow region. Thinning may involve the upper crust, lower crust, or both. Decompression by thinning of thickened crust and lateral flow of deep crust is slow if thinning is restricted to the lower crust, but can be significantly faster if the upper crust is extended, thinned and/or rapidly eroded (Teyssier & Whitney 2002). Because the rate of bulk thinning likely decays with time, this mechanism may initiate decompression, but acting alone it is unlikely to maintain near-isothermal conditions during substantial decompression. Crustal thinning combined with buoyancy may, however, produce the observed and modelled pressure-temperaturetime (P-T-t) paths. Folding /buckling Folding of the lithosphere under compression has been shown to occur in oceanic lithosphere (Wiessel et al. 1980; Gerbault 2000), and possibly also in continental lithosphere (Martinod & Davy 1994; Burg et al. 1994). Buckling theory predicts the amplification of folds of a particular wavelength, given appropriate viscosity contrasts and layer thicknesses. As an antiformal buckle develops (Fig. 2c), a positive feedback relation between folding and erosion may cause an acceleration of fold amplification, resulting in effective, localized lithospheric uplift. This principle has been applied to the exhumation of metamorphic rocks in the Namche-Barwa syntaxis of the eastern Himalayas (Burg et al 1997). The combined effect of fold amplification and erosion may allow exhumation under nearisothermal conditions, such as the P-T path

DECOMPRESSION MELTING AND EXHUMATION

317

Fig. 2. Some of the decompression mechanisms described in the text (a) erosion of uplifted rocks (may be coupled with local faulting/crustal shortening); (b) crustal thinning associated with normal faulting (also shown are schematic 400 and 700 °C isotherms and a partially molten mid-crustal zone); (c) crustal buckling (e.g. at erogenic syntaxes); (d) exhumation of subducted continental crust; and (e) upwelling of a partially molten diapir with associated downflow of denser rocks (upwelling may be driven by density inversion or triggered by removal of the upper crust or thinning of the deep crust).

318

D. L.WHITNEY ET AL

inferred for metamorphic rocks at the NamcheBarwa and Nanga-Parbat syntaxes (Burg & Podladchikov 2000). Buckling may therefore be an effective mechanism of localization of lithospheric uplift. If this lithospheric uplift is accompanied by fast glacial and/or fluvial erosion, heat will be advected upward and a positive feedback relation between crustal softening and strain localization will occur and lead to localized, near-isothermal decompression. Exhumation of ultrahigh-pressure rocks (continental subduction) Continental subduction provides a means of burying and heating a large volume of fertile material that is continually being replenished as long as subduction is active. This mechanism can produce large volumes of melt in orogens, and results in the most dramatic examples of large-magnitude decompression of continental material. Ultrahigh-pressure (UHP) terranes represent continental material that has been subducted to depths of > 100-150 km and exhumed, as shown by the presence of coesite and diamond in metamorphosed supracrustal rocks (Chopin 1984; Coleman & Wang 1995). The largest-magnitude isothermal decompression in continental orogens is recorded by UHP terranes (e.g. Zhang et al 1991 \ Nakamura & Hirajama 2000). UHP terranes now exposed at the surface typically represent crustal slices exhumed at/near the suture zone, and some are in fault contact with blueschist and serpentinite complexes (Dora-Maira, Western Alps; Dabie Shan, China). The subducted crust that returns to the surface as an identifiable UHP terrane may have remained close to the subduction zone and been exhumed via buoyancy-driven

exhumation of a slice bounded below by a thrust and above by a normal fault (Chemenda et al 1995) (Fig. 2d). Subducted continental crust that does not experience a return path near the suture may remain accreted to the base of the overriding plate as slices of continental crust and lithospheric mantle, or may be added to the base of orogenic crust during melting of buoyantly ascending crustal material. Highly oblique subduction, such as proposed for the subducting continental lithosphere under Tibet (Tapponnier et al. 2001), may facilitate prolonged burial/heating and delayed exhumation, creating the right thermal conditions for melting during ascent (Fig. 3). Active collisional orogens provide information about the scale of continental subduction. For example, geophysical studies of subcrustal structure beneath the Himalayan orogen show that a zone of low-density material resides beneath Tibet as well as the western Himalayas (Nelson et al 1996; Van der Voo et al 1999). This zone comprises in part continental crust from the subducted Indian plate (Zhao et al 1993), as well as subducted continental material from the Eurasian plate (Tapponnier et al 2001). In the western Himalayas-Hindu Kush-Pamir zone of central Asia, geophysical evidence documents ongoing steep subduction of continental lithosphere, including low-density crustal rocks (the Indian continental margin). In this region, earthquake hypocentres (Searle et al 2001) and tomographic images based on P-wave velocities (Van der Voo et al 1999) show a steeply dipping region of continental lithosphere that has descended into the mantle to ~300 km depth. To the east, under southern Tibet, Indian lithosphere is currently subducting at a shallow angle (Zhao et al 1993; Owens & Zandt 1997). Evidence from active and ancient orogens

Fig. 3. Tectonic model for continental subduction, decompression and partial melting. Advanced subduction of continental material along multiple, sequential subduction zones (now sutured). Modified from the model of Tapponnier et al (2001). The oldest subducted continental material has risen buoyantly, melted and contributed to crustal thickening of the Tibetan Plateau.

DECOMPRESSION MELTING AND EXHUMATION

suggests that continental subduction is a major process in the evolution of orogens, and that exhumation of subducted continental rocks may be isothermal at high temperatures. The role of melting in this process remains to be discussed (see following section on 'Partial melting in orogens'). Buoyancy / diapirism Lower-density rocks underlying higher-density rocks may rise buoyantly if rheological properties allow flow. During orogeny, a regionalscale density inversion can be achieved in several ways: (1) tectonic stacking of terranes/ thrust slices that are more dense than underlying felsic basement; (2) subduction of continental crust into the mantle; and (3) reduction in bulk density of rocks by addition of felsic magma, with no large-scale segregation of melt. One or more of these situations, combined with the fact that rocks are generally weak at depth owing to elevated temperature and the presence of even small amounts of melt, creates a condition for gravitational instabilities to develop and grow (Fig. 2e). For the case of schist/ gneiss overlying granitic basement, the density contrast (~0.1-0.3 g cm~ 3 ) is sufficient for solid-state diapirism (Fletcher 1972; Soulaet al. 2001). The magnitude of decompression of rising diapirs may be significant, as flow is along subvertical trajectories for a significant portion of the ascent. Once initiated, diapirism may be self-sustained by a positive feedback relation between decompression and melting. The presence of melt lowers the bulk density within the diapir relative to its surroundings, enhancing upward flow and decompression. Rise of partially molten crust is accompanied by downflow of surrounding rocks, which may be transformed to granulite and eclogite that accumulate in the lower crust. The rate of decompression of deep crustal rocks by buoyancy-driven flow will be higher if diapirism is coupled with thinning of thickened crust, particularly if the upper crust is removed by extension, tectonic denudation and/or erosion. The rising diapir may localize upper crustal extension, or the removal of upper crust may drive the buoyant rise of partially molten crust. It is likely that these processes are coupled because they involve a positive feedback relation through the generation of melt (Teyssier & Whitney 2002). The rise of diapirs comprising partially molten crust can account for largemagnitude, near-isothermal decompression if the rate of ascent is fast enough to maintain

319

elevated temperatures and some melt remains in the system.

Partial melting in orogens: mechanisms and consequences Two major processes that characterize collisional orogens are crustal melting and the exhumation of formerly deep continental material. These processes may be genetically linked, as demonstrated by the P-T-t histories of migmatite complexes, and in particular migmatite-cored gneiss domes that are observed in exhumed orogens worldwide. Migmatite diapirs and gneiss domes In orogens, the geological expression of partially molten crust that has experienced high-temperature, near-isothermal decompression is commonly a migmatite-cored gneiss dome. Gneiss domes occur in orogens ranging in age from Archaean through Cenozoic and in tectonic setting from wide (hundreds to thousands of kilometres wide) orogens to narrow (< 10 km wide) shear zones. Well-documented examples occur in the North American Cordillera, Himalayas (Zanskar, Karakorum, Garwhal, South Tibet), Pamirs, Alps (central Alps, Aegean region, Anatolia), Iberian and French Variscides, the Bering Sea region (Alaska, Russia), and the Appalachians, as well as numerous domes in Precambrian terranes. Typically, more than one dome is present in a region, they are elongate and aligned parallel to the strike of the orogen, and they are characterized by a core of anatectic migmatites, orthogneiss and/ or granitoids surrounded by high-grade metasedimentary rocks. In some orogens, there is a characteristic spacing between gneiss domes, e.g. 40-50 km in the northern Cordillera, and 25 + 5 km in the northern Appalachians (Fletcher 1972). The short dimension of gneiss domes ranges from ~4km (Naxos, Greece; Baltimore, USA) to 60 km (Velay, Massif Central, France), but most domes have a short diameter of 15-25 km. The origin of gneiss domes has been debated for more than 50 years (Eskola 1949). Their origin has been ascribed to diapirism (Berner et al 1972; Ramberg 1980; Calvert et al 1999), crustal shortening (Ramsay 1967; Burg et al 1984; Rolland et al 2001), extension (Chen et al 1990; Brun & Van Den Driessche 1994; Escuder Viruete et al 2000), or more complicated models invoking both contraction and extension (Lee et al 2000),

320

D. L. WHITNEY ET AL.

crustal flow/extrusion (Beaumont et al. 2001), or extension-controlled upwelling of partially molten crust (Vanderhaeghe et al. 1999). In recent years, diapirism has been rejected as a general mechanism for the generation of gneiss domes because in many cases the structural doming event is believed to post-date partial melting/magmatism (e.g. Lee et al. 2000; Holland et al 2001). In most orogens, the crustal structure beneath gneiss domes is not known, but in southern Tibet, seismic profiles suggest that the partially molten (high conductivity, low seismic velocity) mid-crust extends to the base of the Kangmar dome (Nelson et al. 1996). Recent papers have suggested links between the deep crust of the southern Tibet and the Himalayan wedge, including flow of ductile Tibetan crust into the Himalayan wedge (Beaumont et al. 2001; Grujic et al. 2002). An example of the relationship between decompression and doming is found in the Shuswap metamorphic core complex, British Columbia, where a series of elongate gneiss domes are aligned along the strike of the belt near its eastern margin. These domes are approximately 15 km across their short axis, are located ~40-50km apart, and include the Frenchman's Cap, Thor-Odin, Pinnacles and Valhalla domes. The high-grade core of the Thor-Odin dome experienced near-isothermal decompression from P > 10 kbartoP < 4 kbar (Norlanderefa/. 2002). The dome is defined in part by the transition from migmatitic rocks that retain a coherent metamorphic layering, to migmatites dominated by the granitic fraction and characterized by lack of a coherent solid framework (Vanderhaeghe etal. 1999). Isothermal decompression of migmatite domes The advection of partially molten material from >30km depth towards the Earth's surface is a significant agent of heat transfer during orogeny. The temperature-time history at shallower levels depends on the final emplacement depth of the diapir and the geothermal gradient of the upper crust: these are controlled by surficial processes and upper crustal deformation (e.g. extension, erosion). Thermal modelling can be used to evaluate the conditions required for near-isothermal decompression and to examine the thermal effects of migmatite diapirs following ascent to the upper crust. We used time-dependent thermal modelling to quantify the relationship between isothermal

decompression of a diapir and cooling rates (Fig. 4). The advection-diffusion equation is solved in two dimensions using an explicit finitedifference method (e.g. Noye 1982). These models illustrate the thermal response of a diapir rising from the deep crust to the middle crust and do not address the mechanical issues of emplacement of diapirs into the rigid upper crust. Despite the mechanical limitations, important information regarding the relationship between pressure-temperature (P-T) and temperature-time (T-f) paths is obtained through this analysis. The predicted T-t paths (Fig. 5) can be compared to observed paths (Fig. 1) to evaluate the exhumation history. The diapir is modelled as a piston 15 km wide with an initial temperature of 775 °C (Fig. 4a). The diapir ascends from a depth of 30 km at a constant rate to mid-crustal levels. Points within the diapir are advected vertically; exhumation rates investigated by the model were 2, 5, 10, 15 and 20 km Ma"1. The more rapid rates are consistent with rates recorded by rocks exposed in migmatite-cored domes (e.g. Brown & Dallmeyer 1996; Calvert et al. 1999). The initial geothermal gradient is non-linear, with an increase in temperature from 320 °C at 15 km to 550 °C at 16 km. Below this interval, which represents the base of a rigid upper crustal lid, the geothermal gradient changes to 10 °C km"1. The results shown in Fig. 4 illustrate the effects of the diapir stopping at the base of the upper crustal lid (Fig. 4b) and piercing the upper crustal lid (Fig. 4c). Pressure-temperature-time paths calculated for various points within the diapir, according to the second model (the diapir pierces the upper crust) and an exhumation rate of 20 km Ma"1, are illustrated in Fig. 5a. These results show that, even at extreme exhumation rates, a rock at the top of the diapir (initial depth = 30 km) loses heat to the surroundings during decompression, and therefore records cooling during decompression. In contrast, rocks within the diapir (initial depth — 35 or 39 km) retain heat for a longer period of time during decompression, resulting in a component of near-isothermal decompression. The numerical experiments therefore predict that rocks within a diapir initiating within the deep crust maintain significantly high T (>700 °C) during decompression to shallow mid-crustal levels. Once the diapir is emplaced at a shallow level, the associated cooling paths predict a range of cooling rates from 24 °C Ma~ ! to as high as 400°CMa~ 1 (Fig. 5b). Both models shown in Fig. 4b and c predict greatest cooling rates of 400°CMa~ 1 during exhumation for rocks at

DECOMPRESSION MELTING AND EXHUMATION

321

Fig. 4. (a) Initial conditions for time-dependent numerical experiments. The initial geothermal gradient is determined using a heat production value of 9.6 x 10~4 jjiW kg~ ! for the upper crust and 5 x 10~5 jjiW kg~ for the lower crust. The boundary between the upper and lower crust is located at 15 km (arrow). The diapir is modelled as a piston, the top of which is initially at 30 km depth and moves vertically at a given rate. rinitial for points within the piston is 775 °C. Boundary conditions are constant basal heat flux and constant surface temperature, Tsurf = 20 °C. (b) Diapir is exhumed to the base of the upper crust at 20 km Ma"1. Dashed contours show position of isotherms after advection (1 Ma); black contours represent thermal structure after 5 Ma of cooling at t = 6 Ma. Contour interval = 200 °C. (c) Diapir is exhumed to a position within the upper crust at a rate of 20 km Ma'1. Contours are the same as in (b). In both models, rocks near the top of the diapir record the greatest cooling, but the magnitude of cooling is greater in the case where the diapir pierces the upper crust.

322

D. L.WHITNEY ET AL.

Fig. 5. (a) Depth (pressure)-temperature diagram showing model results for a particle point located at the top of the diapir (initial depth, Zj = 30 km), a point 5 km below the top of the diapir (Zj = 35 km) and a point 9 km below the top (Zi = 39 km). The decompression rate for all three cases is 20 km Ma"1. The top of the diapir cools continuously during decompression, but the points modelled within the diapir experience near-isothermal decompression until the diapir pierces the upper crust, at which point the diapir experiences isobaric cooling. The grey shaded region illustrates the approximate location of dehydration melting reactions involving biotite. (b) The T—t plot corresponding to the same conditions as in (a) with similar labels and line styles. This T—t diagram illustrates the rapid cooling rates of the modelled processes.

the top of the diapir. Following exhumation, the cooling rate slows to 24 °C Ma"1. The average cooling rate recorded for a rock at the top of the diapir is 56 °C Ma"1. A rock within the diapir, however, records a slower average cooling rate of 40 °C Ma"1. These cooling rates are consistent with observed cooling rates determined by thermochronology of migmatite-cored domes (e.g. Calvert et al 1999; Vanderhaeghe et al 1999). Observed P-T-t paths for migmatite-cored domes suggest rapid isothermal decompression of the deep crust. Calculated P-T-t paths further support buoyancy-driven flow as a viable mechanism for rapid exhumation of these deep rocks. To understand how these domes develop in orogens requires addressing the general question of partial melting during orogenesis. In this chapter thus far, we have emphasized the importance of decompression specifically isothermal decompression - and the tectonic/thermal fate of partially molten crust. We have focused on relatively small(crustal-) scale observations, but will now consider larger- (lithosphere-) scale processes that might contribute to melting of continental crust. By doing so, we can address the fundamental question of how and why the deep crust of orogens attains such high melt fractions over such a great crustal thickness (Nelson et al. 1996; Schilling & Partzsch 2001).

Buoyant return of subducted continental crust The contribution of continental subduction to melting in orogens requires further consideration. In the mid- to deep crust of the southern Tibetan Plateau, it is difficult to explain the presence of the amount of melt inferred (20 vol.%) by simple heating during crustal shortening, as the mantle is not unusually hot beneath this part of Tibet (Nelson et al 1996; Chen et al. 1996). In addition, the rapid rate of underthrusting (subduction) of Indian crust beneath Eurasia (55-60 mm a"1; Patriat & Achache 1984; Le Pichon et al. 1992) and the estimated low temperature at the Indian plate moho at the modern suture (inferred from heat flow data; Gupta 1993) suggest that the lower crust of the subducting Indian continent is at moderate temperatures (<800 °C; Henry et al 1997). Nelson et al (1996) proposed that crustal heat production in 70 km thick continental crust caused partial melting in the Tibetan crust. This mechanism, however, requires the presence of a free aqueous fluid for flux melting, and this combination of variables (enough time for heat production to promote regional-scale melting, the presence of abundant aqueous fluids) is unlikely. However, decompression due to thinning, extension, or buoyant rise of deep crust (including subducted Indian or Eurasian continental material)

DECOMPRESSION MELTING AND EXHUMATION

(Fig. 3) may account for large degrees of melting without the presence of water-rich fluid and without the presence of elevated temperatures (i.e. >800°C). The UHP terranes identified at the Earth's surface represent a small fraction of the total volume of subducted continental material. For example, the amount of Indian continental crust that is estimated to have been buried/subducted beneath the Eurasian plate cannot be accounted for by the amount of crustal thickening in the Himalayas. Some of the missing crust has likely flowed under southern Tibet. Although some subducted continental crust may become dense enough to sink into the mantle as eclogite (Le Pichon et al. 1992), some crust may remain less dense than the surrounding mantle rocks (Hermann 2002) and can rise buoyantly. The question of the fate of subducted continental crust is related to the present discussion because the buoyant rise of subducted continental crust may contribute to partial melting in collisional orogens, by providing source rocks for partial melting during decompression and/or by influencing decompression mechanisms in the overlying (non-subducting) crust. It is unclear at present whether exhumed UHP rocks were partially melted during ultrahighpressure metamorphism. Some UHP terranes contain abundant migmatites (e.g. Dabie Shan), but the general view is that the partial melting occurred in a later event, unrelated to UHP metamorphism (Coleman & Wang 1995). Furthermore, geochemical studies in some exhumed ultrahigh-pressure terranes suggest that the continental material was 'old, cold, and dry' (Coleman & Wang 1995) prior to subduction. Sharp et al. (1993), however, proposed that the stable isotope values of UHP schists from the Dora Maira Massif indicated equilibration with a melt, possibly represented by kyanite + jadeite + garnet + quartz layers in the schists (Schreyer et al 1987). We believe the relationships among UHP metamorphism, exhumation of UHP rocks and partial melting deserve further investigation, but, whether or not exhumed UHP terranes experienced partial melting, it is reasonable to propose that some subducted continental material has sufficient fertility and reaches P-T conditions appropriate for partial melting during deep subduction and/or decompression.

Summary Large-magnitude (>10km), regional-scale decompression that is nearly isothermal at elevated temperatures (>700°C) may be driven

323

by a variety of mechanisms, including surficial and deep crust/mantle processes that may influence each other through feedback relationships. If decompression is a necessary condition for large-scale melting of continental crust, then a significant source of material and/or a major driving force for melting in collisional orogens may be subducted continental material. The transfer of continental material, including melt, from the subducting plate to the non-subducting plate may result in decompression-driven partial melting, with the melt accumulating in the nonsubducting plate. The large volume of melt in a layer within the thickened crust fundamentally changes the balance between tectonic and buoyancy forces in a collisional orogen and results in mechanical decoupling of continental crust and lithospheric mantle. Within continental crust, upper and deep levels may initially be mechanically coupled (e.g. if decompression is driven by surficial or other upper crustal processes), but these too become decoupled through time as partial melting proceeds and the deep crust flows laterally (channel flow) and/or vertically (diapirism). The rise of partially molten crust and associated downflow of denser country rocks signifies decoupling of orogenic crust from the mantle lithosphere and of deep crust from upper crust. We thank Olivier Vanderhaeghe for his comments on the paper, and acknowledge reviews by Mike Brown and J.-L. Vigneresse. In particular, the helpful suggestions of Mike Brown improved the discussion of migmatites and melting relationships. This work was partially supported by NSF grant EAR-9814669.

References AUDREN, C. & TRIBOULET, C. 1993. P-T-t deformation paths recorded by kinzigites during diapirism in the western Variscan belt (Golfe du Morbihan, southern Brittany, France). Journal of Metamorphic Geology, 11, 337-356. BEAUMONT, C., JAMIESON, R.A., NGUYEN, M.H & LEE, B. 2001. Himalayan tectonics explained by extrusion of a low-viscosity crustal channel coupled to focused surface denudation. Nature, 414, 738-742. BERNER, H., RAMBERG, H. & STEPHANSSON, O. 1972. Diapirism in theory and experiment. Tectonophysics, 15, 197-218. BROWN, L.D., ZHAO, W. et al 1996. Bright spots, structure, and magmatism in Southern Tibet from INDEPTH seismic reflection profiling. Science, 274, 1688-1690. BROWN, M. 1979. The tectonic evolution of the Precambrian rocks of the St. Malo region, Armorican Massif, France. Precambrian Research, 6, 1-21. BROWN, M. 1994. The generation, segregation, ascent and emplacement of granite magma; the

324

D. L.WHITNEY ET AL

migmatite-to-crustally-derived granite connection in thickened orogens. Earth Science Reviews, 36, 83-130. BROWN, M. & DALLMEYER, R.D. 1996. Rapid Variscan exhumation and the role of magma in core complex formation; southern Brittany metamorphic belt, France. Journal of Metamorphic Geology, 14, 361-379. BROWN, M. & SOLAR, G.S. 1998. Shear zone systems and melts: feedback relations and self-organization in orogenic belts. Journal of Structural Geology, 20,211-227. BRUN, J.-P. & VAN DEN DRIESSCHE, J. 1994. Extensional gneiss domes and detachment faults - structure and kinematics. Bulletin of the Geological Society of France, 165,519-530. BUCK, W.R., 1988. Flexural rotation of normal faults. Tectonics, 7, 959-973. BURG, J.-P. & PODLADCHIKOV, Y. 2000. From buckling to asymmetric folding of the continental lithosphere; numerical modelling and application to the Himalayan syntaxes. In: KHAN, M.A., TRELOAR, P.J., SEARLE, M.P. & JAN, M.Q. (eds), Tectonics of the Nanga Parbat Syntaxis and the Western Himalaya, Geological Society, London, Special Publications, 170, 219-236. BURG, J.-P., GUIRAUD, M., CHEN, G.M. & Li, G.C. 1984. Himalayan metamorphism and deformations in the north Himalayan Belt (southern Tibet, China). Earth and Planetary Science Letters, 69, 391-400. BURG, J.P., DAVY, P. & MARTINOD, J. 1994. Shortening of analogue models of the continental lithosphere; new hypothesis for the formation of the Tibetan Plateau. Tectonics, 13, 475-483. BURG, J.-P., DAVY, P., NIEVERGELT, P., OBERLI, F., SEWARD, D., DIAO, Z. & MEIER, M. 1997. Exhumation during crustal folding in the NamcheBarwa syntaxis. Terra Nova, 9, 53-56. CALVERT, A., GANS, P.B. & AMATO, J.M. 1999. Diapiric ascent and cooling of a sillimanite gneiss dome revealed by 40Ar/39Ar thermochronology: the Kigluaik Mountains, Seward Peninsula, Alaska: In: RING, U., BRANDON, M.T., LISTER, G.S. & WILLETT, S.D. (eds) Exhumation Processes: Normal Faulting, Ductile Flow, and Erosion. Geological Society, London, Special Publications, 154, 205-232. CHEMENDA, A.I., MATTAUER, M., MALAVIEILLE, J. & BOKUN, A.N. 1995. A mechanism for syncollisional deep rock exhumation and associated normal faulting: results from physical modeling. Earth and Planetary Science Letters, 132, 225-232. CHEN, L., BOOKER, J.R., JONES, A.G., Wu, N., UNSWORTH, M.J., WEI, W. & TAN, H. 1996. Electrically conductive crust in southern Tibet from INDEPTH magnetotelluric surveying. Science, 274, 1694-1696. CHEN, Z., Liu, Y., HODGES, K.V., BURCHFIEL, B.C., ROYDEN, L.H & DENG, C. 1990. The Kangmar Dome: a metamorphic core complex in southern Xizang (Tibet). Science, 250, 1552-1556. CHOPIN, C. 1984. Coesite and pure pyrope in highgrade blueschists of the western Alps: a first

record and some consequences. Contributions to Mineralogy and Petrology, 86, 107—118. CLEMENS, J.D & MAWER, C.K. 1992. Granitic magma transport by fracture propagation. Tectonophysics, 204, 339-360. CLEMENS, J.D. & VIELZEUF, D. 1987. Constraints on melting and magma production in the crust. Earth and Planetary Science Letters, 86, 287-306. COLEMAN, R.G. & WANG, X. 1995. Ultrahigh-Pressure Metamorphism. Cambridge University Press, New York. DUNN, S.R. & MEDARIS, L.D. 1989. Retrograded eclogites in the Western Gneiss region of a portion of the Scandinavian Caledonides. Lithos, 22, 229-245. ENGLAND, P.C. & THOMPSON, A.B. 1984. Pressuretemperature-time paths of regional metamorphism. I. Heat transfer during the evolution of regions of thickened continental crust. Journal of Petrology, 25, 894-928. ENGLAND, P.C. & THOMPSON, A.B. 1986. Some thermal and tectonic models for crustal melting in continental collision zones. In: COWARD, M.P. & RIES, A.C. (eds) Collision Tectonics. Geological Society, London, Special Publications, 19, 83-94. ESCUDER VlRUETE, J., INDARES, A. & ARENAS, R.,

2000. P-T paths derived from garnet growth zoning in an extensional setting: an example from the Tormes Gneiss Dome (Iberian Massif, Spain). Journal of Petrology, 41, 1489-1515. ESKOLA, P.E. 1949. The problem of mantled gneiss domes. Quarterly Journal of the Geological Society, London, 104, 461-476. FAYON, A.K., WHITNEY, D.L. & TEYSSIER, C. 2002. Isothermal decompression of migmatites: diapirism or normal faulting? Geological Society of America Abstracts with Programs, 34, 333. FLETCHER, R.C. 1972. Application of a mathematical model to the emplacement of mantled gneiss domes. American Journal of Science, 272, 197 — 216 GERBAULT, M. 2000. At what stress level is the central Indian Ocean lithosphere buckling? Earth and Planetary Science Letters, 178, 165-181. GRUJIC, D., HOLLISTER, L.S. & PARRISH, R.R. 2002. Himalayan metamorphic sequence as an orogenic channel: insight from Bhutan. Earth and Planetary Science Letters, 198, 177-191. GUPTA, M.L. 1993. Is the Indian shield hotter than other Gondwana shields? Earth and Planetary Science Letters, 115, 275-285. HENRY, P., LE PICHON, X. & GOFFE, B. 1997. Kinematic, thermal, and petrological model of the Himalayas: constraints related to metamorphism within the underthrust Indian crust and topographic elevation. Tectonophysics, 273, 31-56. HERMANN, J. 2002. Experimental constraints on phase relations in subducted continental crust. Contributions to Mineralogy and Petrology, 43, 219-235. HIRN, A., SAPIN, M., LEPINE, J.C., DIAZ, J. & MEI, J. 1997. Increase in melt fraction along a south-north traverse below the Tibetan Plateau: evidence from seismology. Tectonophysics, 273, 17-30.

DECOMPRESSION MELTING AND EXHUMATION HOLLOWAY, J.R. & BLANK, J.G. 1994. Application of experimental results to C-O-H species in natural melts. In: CARROLL, M.R. & HOLLOWAY, J.R. (eds) Volatiles in Magmas. Reviews in Mineralogy, 30, 187-230. INGER, S. & HARRIS, N. 1993. Geochemical constraints on leucogranite magmatism in the Langtang Valley, Nepal, Himalaya. Journal of Petrology, 34, 345-368. JONES, K.A. & BROWN, M. 1990. High-temperature 'clockwise' P-Tpaths and melting in the development of regional migmatites: an example from southern Brittany, France. Journal of Metamorphic Geology,*, 551-578. LEE, J., HACKER, B.R. et al. 2000. Evolution of the Kangmar Dome, southern Tibet: structural, petrologic, and thermochronologic constraints. Tectonics, 19, 872-895. LE FORT, P., CUNEY, M., DENIEL, C., FRANCELANORD, C., SHEPPARD, S.M.F., UPRETTI, B.N. & VIDAL, P. 1987. Crustal generation of the Himalayan leucogranites. Tectonophysics, 134, 39-57. LE PICHON, X., FOURNIER, M. & JOLIVET, L. 1992. Kinematics, topography, shortening, and extrusion in the India-Eurasia collision. Tectonics, 11, 1085-1098. LISTER, G.A. & DAVIS, G.A. 1989. The origin of metamorphic core complexes and detachment faults formed during Tertiary continental extension in the northern Colorado River region, U.S.A. Journal of Structural Geology, 11, 65-94. MARTINOD, J. & DAVY, P. 1994. Periodic instabilities during compression of the lithosphere; 2, Analogue experiments. Journal of Geophysical Research, 99, 12057-12069. McKENZiE, D. 1984. The generation and compaction of partially molten rocks. Journal of Petrology, 25, 713-765. MONTEL, J.M., MARIGNAC, C., BARBEY, P. & PICHAVANT, M. 1992. Thermobarometry and granite genesis: the Hercynian low-P high-T Velay anatectic dome (French Massif Central). Journal of Metamorphic Geology, 10, 1-15. NAKAMURA, D. & HIRAJAMA, T. 2000. Granulitefacies overprinting of ultrahigh-pressure metamorphic rocks, northeastern Su-Lu region, Eastern China. Journal of Petrology, 41, 563-582. NELSON, K.D., ZHAO, W. et al. 1996. Partially molten middle crust beneath southern Tibet: synthesis of Project INDEPTH results. Science, 274, 16841688. NORLANDER, B.N., WHITNEY, D.L., TEYSSIER, C. &

VANDERHAEGHE, O. 2002. Partial melting and decompression of the Thor-Odin dome, Shuswap metamorphic core complex, Canadian Cordillera. Lithos,6l, 103-125. No YE, J. 1982, Finite difference method for partial differential equations. In: NOYE, J. (ed.) Numerical Solutions of Partial Differential Equations. NorthHolland, Amsterdam, 3-137. NYMAN, M.W., PATTISON, D.R.M. & GHENT, E.D. 1995. Melt extraction during formation of K-feldspar + sillimanite migmatites, west of Revel-

325

stoke, British Columbia. Journal of Petrology, 36, 351-372. OWENS, T.J. & ZANDT, G. 1997. Implications of crustal property variations for models of Tibetan Plateau evolution. Nature, 387, 37-53. PATRIAT, P. & ACHACHE, J. 1984. India-Asia collision chronology and implications for crustal shortening and driving mechanisms of plates. Nature, 311,615-621. Pous, J., MUNOZ, J.A., LEDO, J.J. & LIESA, M. 1995. Partial melting of subducted continental lower crust in the Pyrenees. Journal of the Geological Society, London, 152, 217-220. RAMBERG, H. 1980. Diapirism and gravity collapse in the Scandinavian Caledonides. Journal of the Geological Society, London, 137, 261-270. RAMSAY, J.G. 1967. Folding and Fracturing of Rocks. McGraw-Hill, New York. REY, P. 1993. Seismic and tectonometamorphic characters of the lower continental crust in Phanerozoic areas: a consequence of post-thickening extension. Tectonics, 12, 580-590. ROLLAND, Y., MAHEO, G., GUILLOT, S. & PECHER, A. 2001. Tectono-metamorphic evolution of the Karakorum Metamorphic complex (Dassu-Askole area, NE Pakistan): exhumation of mid-crustal HT-MP gneisses in a convergent context. Journal of Metamorphic Geology, 19, 717-737. SAWYER, E.W. 2001. Melt segregation in the continental crust: distribution and movement of melt in anatectic rocks. Journal of Metamorphic Geology, 19, 291-309. SCHILLING, F.R. & PARTZSCH, G.M. 2001. Quantifying partial melt fraction in the crust beneath the central Andes and the Tibetan Plateau. Physics and Chemistry of the Earth, 26, 239-246. SCHREYER, W., MASSONE, H.-J. & CHOPIN, C. 1987. Continental crust subducted to depths near 100 km: implications for magma and fluid genesis in collision zones. In: MYSEN, B.O. (ed.) Magmatic Processes: Physicochemical Principles. Geochemical Society Special Publication, 1, 155-163. SEARLE, M., HACKER, B.R. & BILHAM, R. 2001. The Hindu Kush seismic zone as a paradigm for the creation of ultrahigh-pressure diamond- and coesitebearing continental rocks. Journal of Geology, 109, 143-153. SHARP, Z.D., ESSENE, E.J. & HUNZIKER, J.C. 1993. Stable isotope geochemistry and phase equilibria of coesite-bearing whiteschists, Dora Maira Massif, western Alps. Contributions to Mineralogy and Petrology, 114, 1-12. Soula, J.-C., DEBAT, P., BRUSSET, S., BESSIER, G., CHRISTOPHOUL, F. & DERAMOND, J. 2001. Thrust-related, diapiric, and extensional doming in a frontal orogenic wedge: example of the Montagne Noire, Southern French Hercynian Belt. Journal of Structural Geology, 23, 1677-1699. TAPPONNIER, P., ZHIQIN, X., ROGER, F., MEYER, B., ARNAUD, N., WITTLINGER, G. & JINGSUI, Y. 2001. Oblique stepwise rise and growth of the Tibetan Plateau. Science, 294, 1671-1677. TEYSSER, C. & WHITNEY, D.L. 2002. Gneiss domes and orogeny. Geology, 30, 1139-1142.

326

D. L.WHITNEY ET AL

THOMPSON, A.B. 2000. Clockwise P-T paths for crustal melting and H2O recycling in granite source regions and migmatite terrains. Lithos, 56, 33-35. VANDERHAEGHE, O., TEYSSIER, C. & WYSOCZANSKI, R. 1999. Structural and geochronological constraints on the role of partial melting during the formation of the Shuswap metamorphic core complex at the latitude of the Thor-Odin Dome, British Columbia. Canadian Journal of Earth Sciences, 36, 917-943. VAN DER Voo, R., SPAKMAN, W. & BIJWAARD, H. 1999. Tethyan subducted slabs under India. Earth and Planetary Science Letters, 171, 7-20. VIELZEUF, D. & HOLLOWAY, J.R. 1988. Experimental determination of the fluid-absent melting relations in the pelitic system. Consequences for crustal differentiation. Contributions to Mineralogy and Petrology, 98, 257-276. WANG, Q., ISHIWATARI, A., et al 1993. Coesitebearing granulite retrograded from eclogite in Weihai, eastern China. European Journal of Mineralogy, 5, 141-152. WHITNEY, D.L. 1992. High pressure metamorphism in the western Cordillera of North America: an example from the Skagit Gneiss, North Cascades,

Washington. Journal of Metamorphic Geology, 10,71-85. WHITNEY, D.L. & IRVING, A.J. 1994. Origin of K-poor leucosomes in a metasedimentary migmatite complex by ultrametamorphism, syn-metamorphic magmatism, and subsolidus processes. Lithos, 32, 173-192. WICKHAM, S. 1987. Crustal anatexis and granite petrogenesis during low-pressure regional metamorphism; the Trois Seigneurs Massif, Pyrenees, France. Journal of Petrology, 28, 127-169. WIESSEL, J.K., ANDERSON, R.N. & GELLER, C.A. 1980. Deformation of the Indo-Australian plate, Nature, 87, 284-291. ZEITLER, P.K., KOONS, P.O. et al 2001. Crustal reworking at Nanga Parbat, Pakistan. Metamorphic consequences of thermal-mechanical coupling facilitated by erosion. Tectonics, 20, 712-728. ZHANG, R. Y., Liou, J.G., ERNST, W.G., COLEMAN, R.G., SOBOLEV, N.V. & SHATSKY, V.S. 1997. Metamorphic evolution of diamond-bearing and associated rocks from the Kokchetav Massif, northern Kazakhstan. Journal of Metamorphic Geology, 15, 479-496. ZHAO, W., NELSON, K.D. et al 1993. Deep seismic reflection evidence for continental underthrusting beneath southern Tibet. Nature, 366, 557-559.

Strain-rate-dependent rheology of partially molten rocks J. L. VIGNERESSE1 & J. P. BURG2 1 CREGU, UMR CNRS 7566 G2R, BP 23, F-54501 Vandoeuvre Cedex, France (e-mail: jean-louis. vigneresse@g2r. uhp-nancy.fr) 2 Geologisches Imtitut, ETH-Zentrum, Sonneggstrasse 5, CH-8006 Zurich, Switzerland (e-mail: [email protected]) Abstract: Coupling or decoupling in the lithosphere is often related to the absence or the presence of a layer of partially molten rocks, although the rheology of such rocks remains unsolved. Theoretical arguments correlated with structural observations provide new insights into the rheology of partially molten rocks, namely migmatites and magma. These rocks are simplified to two-phase pseudo-fluids constituted of a quasi-solid matrix and a variable amount of melt. Previous experiments indicate that the matrix deforms plastically according to a power law. The melt is Newtonian and weakens at high shear strain rates. Because of the heterogeneous distribution of matrix and melt phases, their rheologies cannot be averaged to obtain the rock rheology. Four behaviours are identified, (i) At high stress and strain rates, the viscosity contrast between melt and matrix is lowest. Both phases can accommodate strain at a comparable rate, allowing migmatite and magma bodies to deform as quasi-solid units, (ii) At low strain rates, the viscosity contrast between melt and matrix is highest. Melt deforms and relaxes much faster than the matrix. The simultaneous coexistence of a weak and a strong phase is expressed in a 3D viscosity-strain rate-melt fraction diagram, in which a cusp-shaped surface represents viscosity. The cusp graphically shows that the viscosity of partially molten rocks may jump several orders of magnitude. These jumps, leading to sudden melt segregation, are temporally erratic, (iii) At low strain rates, strain partitioning may lead to internal instabilities and segregation between melt and restitic phases, as observed in the leucosome/melanosome separation, (iv) Cyclic processes follow hysteresis loops and trigger strain localization.

Vertical or horizontal decoupling in the lithosphere requires that there is an interface whose rheology restricts stress transmission. This may occur along a discontinuity, as a fault in the brittle crust, or as weak shear zones in the ductile crust. Weakness is commonly attributed to a compositional change, including soft minerals (Jordan 1987), or to the onset of melting, the molten rocks providing the decoupling interface (Dewey 1988; Block & Royden 1990; see recent review by Vanderhaeghe & Teyssier 2001). At a smaller scale, coupling and decoupling are coeval in migmatites, which were partially molten crustal rocks (Mehnert 1968; Ashworth 1985; Brown 1994). The presence of melt induces strain partitioning and decoupling between the melt and its matrix (Vigneresse & Tikoff 1999). In contrast, large-scale structures of a migmatite body are concordant with those of the surrounding rocks, manifesting some kind of coupling. The rheology of partially molten rocks (PMR) has been approached from a theoretical point of view, importing into Earth sciences observations

and experiments from other disciplines such as soil science, food engineering, polymer rheology and chemical engineering. Crystallizing magma, and migmatites also, consist of solid crystals suspended in a fluid melt. They can both be considered Theologically as two-phase materials at the time of crystallization and melting, respectively. The present chapter identifies rheological behaviours and physical effects to be taken into account for investigating magma segregation and extraction. We first review results concerning the rheology of two-phase materials. PMR are considered to be viscous, with viscosity values rapidly changing from that of the strong to that of the weak phase (Arzi 1978; Barboza & Bergantz 1998; Renner et al 2000). With such considerations, melting and crystallization present symmetrical rheological properties, which is not supported by microstructural observations (Vigneresse et al 1996; Rosenberg 2001). To explore the many aspects of PMR rheology, we extend a previous model that included non-linear interactions between melting, strain partitioning and rheology (Burg & Vigneresse

From: GROCOTT, J., MCCAFFREY, K. J. W., TAYLOR, G. & TIKOFF, B. (eds) 2004. Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 327-336. 0305-8719/04/$15 © The Geological Society of London 2004.

328

J. L. VIGNERESSE & J. P. BURG

2002) and obtain catastrophic events and memory effects due to the hysteretic behaviour of PMR. We first present the intrinsic rheological responses of each phase. We then briefly discuss the limits of the averaging formulation. We attempt to reach a global formulation, which is highly strain-rate-dependent. The ensuing metastable states describe local instabilities. We finally discuss geological implications, but point out that these findings are only in a preliminary, mostly conceptual, stage.

Rheology of two-phase materials Basic rheological laws Rheology is the science of deformation and flow (e.g. Adler et al 1990). In elastic materials, strain disappears as stress is removed. In viscous materials, the strain rate (e) depends in a complex manner on the applied stress (
Very high-temperature rocks produce magma that behaves as a low-viscosity (104 to 1010 Pa s) Newtonian body (Shaw 1972; Holtz et al 1996; Clemens & Petford 1999). The strain rate of rocks deforming by creep is related to stress (T according to a power law:

in which A is a grain-size-sensitive constant, Q is the activation energy and n is the coefficient of the power law, which are experimentally determined. R is the ideal gas constant and T is the absolute temperature. In most rocks, the powerlaw coefficient (n) is close to 3 (Kirby & Kronenberg 1987). In PMR, a weak (the melt) and a strong phase (the matrix) simultaneously coexist. The solid matrix deforms plastically following a flow law close to equation (2). In contrast, the melt is Newtonian at low shear rate (Spera et al. 1988; Kohlstedt et al. 2000) and shows shear weakening when s > 10~4 5 s"1 (Webb & Dingwell 1990; Dingwell et al. 1996). In our model, the maximum strain rate is set to 1 s"1, equivalent to seismic strain, an estimated upper bound for strain-rate values (Scholz 1990; Reimold 1995). In all cases, strain concentrates into the melt (Vigneresse & Tikoff 1999). Complex nonlinear interactions occur, inducing feedback loops between melt production, strain response

and material rheology, and instabilities are a function of the melt fraction (Burg & Vigneresse 2002). In order to investigate in more detail the rheology of melt plus matrix systems, we map the stress-strain rate field in which PMR plot. Equations (1) and (2) are better formulated in logarithmic coordinates, thus allowing direct comparison of the two laws: for the melt and

for the matrix In this formulation, log rj — 1 corresponds to 107 Pa s, the Newtonian viscosity of a granitic melt (Spera et al. 1988; Clemens & Petford 1999). The C(7) term in equation (4) includes the coefficient A of equation (2) and a temperature-dependent term that includes the activation energy Q. We use experimental data with log A = -4.89 and Q = 243 kJ moF1 (Wilks & Carter 1990). We take 800 °C as the approximate temperature for the onset of dehydration melting of biotite (Patino Douce & Beard 1995). C(T) is then set to a constant value of — 32. In the range of strain rates 1Q-4.5 < g < io°s~ 1 , we let viscosity decrease with strain rate (Dingwell et al. 1996). In that range of strain rates, a power law in which n — 2 best matches experimental data. Those values adopted must be considered as average assessments. Departures due to changes in melt composition or temperature (±100°C) do not strongly change the parameters. They vary by about ± 1 logarithmic unit in viscosity, a factor of 10. In all cases this does not alter the shape of the stress-strain rate diagram (Fig. 1). In Figure 1, the melt and the matrix curves intersect at a point with coordinates (12.5, 5.5) if no shear weakening occurs. With shear weakening, the point is located at (15, 12). Both intersections occur at stress and strain-rate magnitudes that are not significant in geological conditions. However, intersection means that, for the corresponding parameters, the viscosity of the matrix is similar to that of the melt. Bracketing geological stress between 0.1 and lOOMPa (5 to 8 in log coordinates), the two lines do not coincide and delimit an irregular pentagon, which is the region that contains the rheology of natural PMR in a log-log diagram. In Cartesian coordinates, this region turns into a triangle because the strain rate remains very low (10~17 to 10° s"1) compared to the stress range (105 to 108 Pa). Within this triangle, the couple (cr, e) varies non-linearly with the melt content.

RHEOLOGY OF A TWO-PHASE MATERIAL

329

Adopting equation (5) or equation (6) is not trivial and has implications for the theoretical behaviour of each phase, because the two descriptions correspond to specific boundary conditions. The arithmetic description corresponds to the same strain rate applied to both phases. In an elasto-viscous PMR, the arithmetic mean viscosity cannot fit the fluid response if the elastic component is significant. It should be selected for a solid-like behaviour in which small deformations are damped. This is the case in a crystallizing magma. Conversely, the geometric formulation supposes that the same stress applies to both phases, which respond as a fluid to a steady or stationary state. The fluidlike behaviour allows the accommodation of large strain with partial elastic recovery. It applies to melting migmatites. The rheological equation of a crystallizing magma can thus be expressed as Fig. 1. Log-log stress-strain rate (cr-e) diagram displaying the relation (3) for a Newtonian melt (n = 1), which turns to shear thinning for e > 10~45 s"1, and the power law (4) (n = 3) for the matrix (see references in the text). Grey lines indicate the corresponding viscosity values.

Hence, we must now investigate the effects of the melt fraction. Rheology of a two-phase pseudo-fluid The rheology of PMR implies introducing a proportion of solid phase into the melt, i.e. solid fraction () relative to melt; O ranges from 0 in the fluid-like case to 1 in the solid-like case. The simplest formulation of bulk rheology consists in estimating either the average strain rate or the average stress that simultaneously applies to both phases. They form the extreme ideal formulation of a mixture. Owing to the large viscosity contrast, averaging is not particularly useful for PMR. The arithmetic and geometric averaging equations are the simplest formulations (Gueguen & Palciauskas 1992). A system composed of two materials with viscosities 77! and 172 has an arithmetic (or Voigt) mean viscosity j]a given by

and a geometric (or Reuss) mean viscosity j]g given by

which can also be written as

This is the general Herschel-Bulkley equation (Adler et al 1990). For melting migmatites, the rheological equation can be expressed as

or alternatively as

This is the general equation for thixotropy (Adler et al. 1990; Barnes 1997). Suffixes 1 and 2 correspond to melt and matrix, respectively. The corresponding viscosity values are plotted with respect to the solid fraction (Fig. 2). The viscosity contrast between the two phases is varied from 101 to 109 Pa s to emphasize differences between the two averaging methods. The arithmetic mean favours the largest values of the average at the expense of small ones. Conversely, the geometric average is damped by the small values of the average. In other words, the arithmetic average favours the strong phase, whilst the geometrical average gives much more weight to the weak phase. We conclude that there is no correct way to address the problem of a heterogeneously

330

J. L. VIGNERESSE & J. P. BURG

Fig. 2. Viscosity contrast between the liquid and solid phases as a function of the solid fraction (4>) for a PMR. Different values of the viscosity contrast (102, 105, 108, 1011 bottom to top) are presented. For each, two curves correspond to the arithmetic (r/a) and geometric (rjg) averages, respectively. They define a system that develops at constant stress and constant strain rate, respectively. The geometric average does not show many differences between a contrast of 102 and 1011 in the end-member values.

varying phase proportion. Any model for averaging rheological properties, whether arithmetic or geometric, fails to represent adequately the rheology of PMR. Averaging formulations involve either stress or strain-rate boundary conditions that apply simultaneously to both phases. Hence, a unique formulation cannot indifferently describe melting and crystallization that proceed from different boundary conditions. A second consequence of averaging methods is that the resulting equations for bulk flow involve both melt and matrix flow. The flowing material is a pseudo-fluid. This may be the case when the viscosity contrast is restricted to 1-3 orders of magnitude. In the case of PMR, the viscosity contrast is huge, up to 15 orders of magnitude, and varies with the imposed strain rate. A limit to the flow would be a Darcy flow, in which the matrix is fixed, which is obviously not a correct model for PMR. Therefore, we suggest another way to examine the rheology of a two-phase material. Rheology of a two-phase material Representing the PMR rheology implies introducing an amount of solid fraction as the third dimension in a (cr, s, <3>) diagram. The applied stress is difficult to estimate. Therefore, we prefer to display the rheology in a diagram scaled with viscosity (Fig. 3). Several considerations are prerequisites.

1 The PMR viscosity is constrained by experimental measurements (Lejeune & Richet 1995; Renner et al 2000) and numerical applications (Vigneresse et al. 1996; Barboza & Bergantz 1998). They demonstrate the increase of viscosity with particle concentration, and thresholds that apply. 2 A 3D diagram must take into account the non-linear effects discussed earlier (Burg & Vigneresse 2002). 3 The bulk melt content is unsatisfactory to define a PMR. We must consider the coexistence of two phases with contrasting rheology. 4 The 3D representation should also coincide with the known viscosity dependence on the solid fraction (Lejeune & Richet 1995) and the solid-like behaviour of concentrated suspensions (Rogers et al. 1994). 5 The model should simultaneously describe the small- and large-scale behaviour of PMR, which correspond to short and long relaxation times of melt and matrix, respectively. The irregular pentagon of Fig. 1 turns, in the 3D diagram, into a cusped viscosity surface integrating the coexistence of the two phases (Fig. 3). In consequence, the strain response to stress, or viscosity determination, is not a single parameter. It varies strongly with applied stress, shear rate or solid fraction. This first-order conclusion strongly contrasts with the equivalent viscosity that would apply for a given solid fraction, independently of strain rate (Arzi 1978). Rheologies for low (^> < 10%) and high ( > 60%) solid fractions are both easy to estimate. Such PMR behave almost exactly as the solid and fluid end-members, with non-linear modifications of the effective viscosity according to the Einstein-Roscoe model of suspensions (reviews in Adler et al. 1990; Liu & Masliyah 1996; Vigneresse et al. 1996). For intermediate solid fractions (10-60%), complexity due to non-linear interactions occurs (Burg & Vigneresse 2002). We first examine the case of low strain rates. The viscosity contrast between melt and matrix is highest (Fig. 3). The relaxation time of the melt is much shorter than for the matrix. This difference generates an instability expressed by the cusp shape of the surface representing viscosity (Fig. 3). The coexistence of the two phases over a range of percentages (0.20 < < 0.75) results in a metastable state. For a given solid fraction, three viscosity values are theoretically possible. The highest and the lowest correspond to stable states and reflect

RHEOLOGY OF A TWO-PHASE MATERIAL

331

Fig. 3. Diagram for strain rate-viscosity-solid fraction (e-Tj-O) that relates strain rate and viscosity according to the proportion of one phase.

the viscosity of the matrix and the melt, respectively. The third, intermediate one is unstable. It reflects the response of a PMR to stress and may fluctuate from the viscosity of the matrix to that of the melt. The fluctuation is rapid, similar to catastrophes as formulated in mathematics (Zeeman 1976; Thorn 1990; Wassermann et al 1992). At higher strain rates, the viscosity contrast between melt and matrix is a few orders of magnitude (Fig. 1). The relaxation times of both phases allow accommodation of their relative motions and there is no instability. The difference in relaxation times between melt and matrix has a second consequence expressed by hysteresis in a stress-strain ratesolid fraction diagram (Fig. 3). The metastable state develops over 0.20 < < 0.75, as stated above. However, it also develops for a fixed percentage of one phase, but at varying strain-rate values. For example, at 70% of melt, and low strain rate (Fig. 3), the sequentially loaded and unloaded PMR may suffer cyclic deformation, but the strong phase has no time to relax completely. Strain adds up at each cycle, resulting in large localization of strain in the weak phase, like it has been experimented on water-saturated sand and clay (Bardet 1995, 1996). In a similar way, experiments on partially saturated sediments,

which have a non-linear rheology, report a drastic increase of strain amplitude while the material is submitted to low-frequency stress cycling (Guyer et al 1995; Tutuncu et al 1998).

Geological implications The levels of stress and strain rate differ when considering the bulk behaviour of a PMR or only local melt segregation. In the following, we discuss some geological implications of our model. Bulk response of PMR to tectonic stresses (high strain rate) PMR is of regional extent (103-105 m) and the ambient stresses are tectonic. We first address melting metatexites, which are low-melt (1020%) migmatites. They commonly form under 400-800 MPa and 750-800 °C (Brown 1994). Metamorphic pressures correspond to the 1025 km depth of the brittle-ductile transition where fracturing occurs at 0.8 to 4 times the vertical load (Byerlee 1978): 100-400 MPa at 1012 km depth, depending on whether tectonics are extensive or compressive. Accordingly, we may assume an ambient differential stress of about 108 Pa (8 in the log-log diagram, Fig. 1). The

332

J. L. VIGNERESSE & J. P. BURG

corresponding strain rate is in the range of 1 to —3 in log scale, with viscosity of 10 7 Pas for the melt and 1010 Pa s for the matrix. The corresponding viscosity contrast is 103 (Fig. 4). The response to ambient tectonic stress is thus mostly the solid-state behaviour of the matrix. One may thus expect to observe structures that are similar to those of the surrounding nonmolten rocks (Nzenti et al 1988; Brun & Bale 1990). High tectonic stresses may impose correspondingly high strain rates onto a granitic body with 40-50% crystals. Although the magma is still rich in melt, it may react as a solid, up to fracturing (Dingwell 1997). At moderate strain rate, plastic deformation locally re-orientates the crystalline fabrics, inducing proto-faults as described in the Mono Creek massif (Saint Blanquat et al 1998). The bulk reaction of a two-phase system to high stress level is common in geology. At very fast strain rates, as during earthquakes, saturated sediments that are ordinarily very soft and without cohesion behave en masse during soil liquefaction (Ishihara 1993).

Bulk response of PMR to low strain rates PMR has similar regional extent as previously (103-105m), but the deformation rate of 10^ 16 s -1 is below the usual strain rates in geology (Pfiffner & Ramsay 1982). This situation applies to the lower continental crust, in which some weakness may form, either from grain size reduction, leading to strain localization, or from partial melting, that may induce magmatism. Such situations are inferred in lower crustal sections presently brought to the surface as in Kohistan (Arbaret et al. 2000), or deduced from geophysical studies as under the Andes (Schilling et al. 1997), the Pyrenees (Partzsch et al. 2000) and the Tibetan Plateau (Hauck et al. 1998). The amount of weak phase does not need to be high. Estimates of 4% melting are provided in the Pyrenees and the Tibetan Plateau (Partzsch et al. 2000). Nevertheless, connection of any melt is assumed from electrical conductivity, which suffices for strain partitioning. In such cases, decoupling can be assumed to develop at very low strain rate.

Fig. 4. Same diagram as Figure 3, but on which we have located: the specific cusp shape instability for vein segregation; the quasi-continuous variation of viscosity that rules the deformation of the bulk massif at low strain rate; the instability due to viscosity variation leading to melanosome segregation from the leucosome; and finally, the zone where hysteresis takes place.

RHEOLOGY OF A TWO-PHASE MATERIAL

Melt segregation at outcrop scale At the mesoscopic scale (1-10 m), low stress is added to the background field and results from local readjustments after en masse motion of melt and matrix. The large viscosity contrast between melt (107 Pa s) and matrix (10 13 Pas) implies a large strain-rate contrast. Strain partitions into the melt and relaxes much faster than in the surrounding matrix. In consequence, the rheological contrast between melt and matrix accentuates at low strain rate, leading to instability, whatever the melt percentage. If stress cycles occur while melt and matrix deform with the same strain rate, sudden jumps between the liquid-like and solid-like rheology, or vice versa, correspond to sudden expulsion of the melt. The low relaxation rate of the matrix makes the periods of melt segregation irregular and discontinuous in time. Irregular bursts of melt segregation have been obtained in numerical (Vigneresse & Burg 2000; Rabinowicz et al 2001) and analogue (Rosenberg & Handy 2001; Barraud et al. 2001) modelling. Conversely, deformation involving the same stress on melt and matrix leads to strain localization (Fig. 5). Repeated loading and unloading increases and amplifies strain localization, resulting in thin shear bands in which a small amount of melt may have focused deformation. With further deformation, the weaker shear bands may be reactivated, showing very large finite strain (see photos in Crawford et al. 1998). Melt segregation within a vein At smaller scale (10~2-1 m), strain partitioning and strain rate are high in the melt because its relaxation time is short. We do not wish to

Fig. 5. Instability developing at quasi-constant stress and leading to strain localization.

333

enter the controversy about the actual proportion of molten and restitic phases and the effects of backreactions changing this proportion (Kriegsman 2001). We simply assume that the flowing melt is a suspension of crystals in a fluid. In suspensions, the velocity gradient and strain partitioning produce a plug-like concentration of solid particles in the centre of a channelled flow, known as the Bagnold effect (Bagnold 1954; Komar 1976). It is common in dykes and in crystal-rich lava flows (Correa-Gomes et al. 2001). In contrast, a concentration of restitic minerals (melanosome) on both sides of channelled melt (leucosome) is commonly described in migmatites. The flat shape of restitic biotites is generally parallel to leucosomes and flow direction (McLellan 1988). This orientation suggests that those minerals have passively rotated and side melanosomes seem to correspond to outward concentration of solid particles, more or less symmetric with respect to the central flow plane. Segregation towards the borders of the flow, as suggested by melanosome-bounded leucosomes in migmatites, thus differs from solid-phase segregation in dykes. It is due to both strain partitioning and chemical effects (Olmsted 1999; Tanaka 2000). Strain partitioning concentrates each phase along the flow direction and perpendicular to vorticity (Fig. 6). Exchange of elements between biotite and microcline (Olsen 1977) added to the mechanical effect could explain why biotites concentrate along each side of the leucosome, with their flat face aligned parallel to the flow direction (McLellan 1983). The concentration of restitic minerals on

Fig. 6. Catastrophic jumps in viscosity that take place at constant low strain rate, between places of different melt content. They lead to segregation of the melanosome either side of leucosomes, as occurs during spinodal decomposition.

334

J. L. VIGNERESSE & J. P. BURG

both sides of the leucosome would therefore be a consequence of instability under a common strain rate. Viscosity determination for PMR A direct consequence of time- or strain-ratedependent rheology is the non-unique determination of an equivalent viscosity for a PMR, given its proportion of melt. More generally, a two-phase material cannot be represented by one single equivalent viscosity. Extending the discussion a step further would be to link the viscosity determination to an underlying mechanism. Power-law rheology classically corresponds to dislocation creep (Kirby & Kronenberg 1987). Diffusion creep corresponds to Newtonian flow. However, recent experiments show that a transition from dislocation creep to diffusion creep may develop when a small fraction of melt is added to an aggregate of solid grains (Kohlstedt et al. 2000). A specific 'granular flow' has been coined for this situation (Paterson 2001). It manifests by a prompt switch from power-law (n — 3) to linear (n = 1) rheology (Kohlstedt et al 2000), which is the situation our model describes. Conclusions We imported models from other disciplines to approach the rheology of partially molten rocks (melting migmatites and crystallizing magmas). We point out that PMR is a material in which it is impossible to assign a given phase proportion at any point. Therefore, average laws fail to represent PMR rheology correctly. To alleviate the problem, we first mapped the PMR rheology within a stress-strain rate diagram in which we could incorporate the phase content. Results show hysteresis and catastrophic jumps of viscosity due to the coexistence of a solid and a fluid phase. Several responses may develop, depending on whether the system responds under high or low strain rate or stress level. We infer that a migmatite body and plutons deform as solid-like mechanical units under high tectonic stress level. At a smaller scale, and under low stress level, instabilities develop under steady strain rate or quasi-constant stress and produce strain localization. Since instabilities involve hysteresis, cyclic loading and unloading may lead to large strain concentrated in small regions. If the same strain rate is imposed on fluid and solid phases, melt segregation and increased vorticity cause veins transecting layering and flow-induced mineral

separation such as restitic minerals aside leucosomes. This paper resulted from numerous and enriching discussions with many colleagues. It also came out from several discussion meetings at Zurich, for which CREGU founding is acknowledged. C. Rosenberg and F. Spera provided quite thoughtful comments.

References ABLER, P., NADIM, A. & BRENNER, H. 1990. Rheological models of suspensions. Advances in Chemical Engineering, 15, 1-67. ARB ARET, L., BURG, J.P., ZEILINGER, G., CHAUDHRY, N., HUSSAIN, S. & DAWWOD, H. 2000. Pre-collisional anastomosing shear zones in the Kohistan arc, NW Pakistan. In: KAHN, M.A., TRELOAR, P.J., SEARLE, M.P. & JAN, M.Q. (eds) Tectonics of the Nanga Parbat Syntaxis and the Western Himalaya. Geological Society, London, Special Publications 170,295-311. ARZI, A A. 1978. Critical phenomena in the rheology of partially melted rocks. Tectonophysics, 44, 173-184. ASHWORTH, J.R. 1985. Migmatites. Blackie & Sons, Glasgow. BAGNOLD, R.A. 1954. Experiments on a gravity-free dispersion of large solid spheres in a Newtonian fluid under shear. Royal Society of London Proceedings, A225, 49-63. BARBOZA, S.A. & BERGANTZ, G.W. 1998. Rheological transitions and the progress of melting of crustal rocks. Earth and Planetary Science Letters, 158, 19-29. BARDET, J.P. 1995. Scaled memory model for undrained behavior of anisotropic clays. Journal of Geotechnical Engineering, 121, 755—765. BARDET, J.P. 1996. Scaled memory description of hysteretic material behavior. Journal of Applied Mechanics, 63, 750-757. BARNES, H.A. 1997. Thixotropy - a review. Journal of Non-Newtonian Fluid Mechanics, 70, 1-33. BARRAUD, J., GARDIEN, V., ALLEMAND, P. & GRANDJEAN, P. 2001. Analog modelling of melt segregation and migration during deformation. Physics and Chemistry of the Earth, A26, 317-323. BLOCK, L. & ROYDEN, L.H. 1990. Core complex geometries and regional scale flow in the lower crust. Tectonics, 9, 557-567. BROWN, M. 1994. The generation, segregation, ascent and emplacement of granite magma: the migmatite-to-crustally-derived granite connection in thickened orogens. Earth Science Reviews, 36, 83-130. BRUN, J.P. & BALE, P. 1990. Cadomian tectonics in northern Brittany. In: D'LEMOS, R.S., STRACHAN, R.A. & TOPLEY, C.G. (eds) The Cadomian Orogeny. Geological Society, London, Special Publications, 51,95-114. BURG, J.P. & VIGNERESSE, J.L. 2002. Non-linear feedback loops in the rheology of cooling-crystallising

RHEOLOGY OF A TWO-PHASE MATERIAL felsic magma and heating-melting felsic rock. In: DRURY, M, DE MEER, S., DE BRESSER, J.H. & PENNOCK, G. (eds) Deformation Mechanisms, Rheology and Tectonics. Geological Society, London, Special Publications, 200, 275-292. BYERLEE, J.D. 1978. Friction of rocks. Pure and Applied Geophysics, 116, 615-626. CLEMENS, J.D. & PETFORD, N. 1999. Granitic melt viscosity and silicic magma dynamics in contrasting tectonic settings. Journal of the Geological Society, London, 156, 1057-1060. CORREA-GOMES,

L.C.,

SOUZA

FlHLO,

C.R.,

MARTINS, C.J.F.N. & OLIVEIRA, E.P. 2001. Development of symmetrical and asymmetrical fabrics in sheet-like igneous bodies: the role of magma flow and wall-rock displacements in theoretical and natural cases. Journal of Structural Geology, 23, 1415-1428. CRAWFORD, M.L., KLEPEIS, K.A., GEHRELS, G. & ISACHSEN, C. 1998. Batholith emplacement at mid-crustal levels and its exhumation within an oblique convergent margin. Tectonophysics, 312, 57-78. DEWEY, J.F. 1988. Extensional collapse of orogens. Tectonics,!, 1123-1139. DINGWELL, D.B. 1997. The brittle-ductile transition in high-level granitic magmas: material constraints. Journal of Petrology, 38, 1635-1644. DINGWELL, D.B., HESS, K.U. & KNOCHE, R. 1996. Granite and granitic pegmatite melts: volumes and viscosities. Transactions of the Royal Society of Edinburgh, Earth Sciences, 87, 65-72. GUEGUEN, Y. & PALCIAUSKAS, V. 1992. Introduction a la Physique des Roches. Hermann, Paris. GUYER, R.A., MCCALL, K.R. & BOITNOTT, G.N. 1995. Hysteresis, discrete memory, and nonlinear wave propagation in rock: a new paradigm. Physical Review Letters, 74, 3491-3494. HAUCK, M.L., NELSON, K.D., BROWN, L.D., ZHAO, W. & Ross, A, 1998. Crustal structure of the Himalayan orogen at -90° east longitude from INDEPTH reflection profiles. Tectonics, 17, 481-500.

HOLTZ, F., SCAILLET, B., BEHRENS, H., SCHULZE, F. &

PICHAVANT, M. 1996. Water content of felsic melts: applications to the rheological properties of granitic magmas. Transactions of the Royal Society of Edinburgh, Earth Sciences, 87, 57—64. ISHIHARA, K. 1993. Liquefaction and flow failure during earthquakes. Geotechnique, 43, 351-415. JORDAN, P.J. 1987. The deformation behaviour of bimineralic limestone-halite aggregates. Tectonophysics, 135, 185-197. KIRBY, S.H. & KRONENBERG, A.K. 1987. Rheology of the lithosphere: selected topics. Reviews of Geophysics, 25, 1219-1244. KOHLSTEDT, D.L., BAI, Q., WANG, Z.C. & MAI, S.

2000. Rheology of partially molten rocks. In: BAGDASSAROV, N., LAPORTE, D. & THOMPSON, A.B. (eds) Physics and Chemistry of Partially Molten Rocks. Kluwer, Dordrecht, 3-28. KOMAR, P.D. 1976. Phenocryst interactions and the velocity profile of magma flowing through dikes and sills. Geological Society of America Bulletin, 87, 1336-1342.

335

KRIEGSMAN, L.M. 2001. Partial melting, partial melt extraction and partial back reaction in anatectic migmatites. Lithos, 56, 75—96. LEJEUNE, A.M. & RICHET, P. 1995. Rheology of crystal-bearing silicate melts: an experimental study of high viscosities. Journal of Geophysical Research, B100, 4215-4229. Liu, S. & MASLIYAH, J.H. 1996. Rheology of suspensions. In: SCHRAMM, L.L. (ed.) Suspensions Fundamentals and Applications in Petroleum Industry. Advances in Chemistry Series 251. American Chemical Society, Washington, DC, 107-176. MCLELLAN, E.L. 1983. Contrasting textures in metamorphic and anatectic migmatites: an example from the Scottish Caledonides. Journal of Metamorphic Geology, 1, 241-262. MCLELLAN, E.L. 1988. Migmatite structures in the Central Gneiss Complex, Boca de Quadra, Alaska. Journal of Metamorphic Geology, 6, 517-542. MEHNERT, K.R. 1968. Migmatites and the Origin of Granitic Rocks. Elsevier, Amsterdam. NZENTI, J.P., BARBEY, P., MACAUDIERE, J. & SOBA, D. 1988. Origin and evolution of the late Precambrian high grade Yaounde gneiss (Cameroon). Precambrian Research, 38, 91-109. OLMSTED, P.D. 1999. Two-state shear diagrams for complex fluids in shear flow. Europhysics Letters, 48, 339-345. OLSEN, S.N. 1977. Origin of the Baltimore gneiss migmatites at Piney Creek, Maryland. Geological Society of America Bulletin, 88, 1089-1101. PARTZSCH, G., SCHILLING, F.R. & ARNDT, J. 2000. The influence of partial melting on the electrical behavior of crustal rocks: laboratory examinations, model calculations and geological interpretations. Tectonophysics, 317, 189-203. PATERSON, M.S. 2001. A granular flow theory for the deformation of partially molten rock. Tectonophysics, 335, 51 -61. PATINO DOUCE, A.E. & BEARD, J.S. 1995. Dehydration-melting of biotite gneiss and quartz amphibolite from 3 to 15 kbar. Journal of Petrology, 36, 707-738. PFIFFNER, O.A., & RAMSAY, J.G. 1982. Constraints on geologic strain rates: arguments from finite strain states of naturally deformed rocks. Journal of Geophysical Research, 87, 311-321. RABINOWICZ, M., GENTHON, P., CEULENEER, G. & HILAIRET, M. 2001. Compaction in a mantle mush with high melt concentrations and the generation of magma chambers. Earth and Planetary Science Letters, 188, 313-328. REIMOLD, W.U. 1995. Pseudotachylite in impact structures - generation by friction melting and shock brecciation?: A review and discussion. Earth Science Reviews, 39, 247-265. RENNER, J., EVANS, B. & HIRTH, G. 2000. On the Theologically critical melt fraction. Earth and Planetary Science Letters, 181, 585-594. ROGERS, C.D.F., DIJKSTRA, T.A. & SMALLEY, T.A. 1994. Particle packing from an Earth science viewpoint. Earth Science Reviews, 36, 59—82.

336

J. L. VIGNERESSE & J. P. BURG

ROSENBERG, C.L. 2001. Deformation of partiallymolten granite: a review and comparison of experimental and natural case studies. International Journal of Earth Sciences, 90, 60-76. ROSENBERG, C.L. & HANDY, MR. 2001. Mechanisms and orientation of melt segregation paths during pure shearing of a partially molten rock analog (norcamphor-benzamide). Journal of Structural Geology, 23, 1917-1932. SAINT BLANQUAT, M. DE, TIKOFF, B., TEYSSIER, C. & VIGNERESSE, J.L. 1998. Transpressional kinematics and magmatic arcs. In: HOLDSWORTH, R.E., STRACHAN, R.A. & DEWEY, J.F. (eds) Continental Transpressional and Transtensional tectonics. Geological Society, London, Special Publications, 135, 327-340. SCHILLING, F.R., PARTZSCH, G., BRASSE, H. & SCHWARZ, G. 1997. Partial melting below the magmatic arc in the central Andes deduced from geoelectromagnetic field experiments and laboratory data. Physics of the Earth and Planetary Interiors, 103, 17-31. SCHOLZ, C.H. 1990. The Mechanics of Earthquakes and Faulting. Cambridge University Press, New York. SHAW, H.R. 1972. Viscosities of magmatic liquids: an empirical method of prediction. American Journal of Science, 272, 870-893. SPERA, F.J., BORGIA, A., STRIMPLE, J. & FEIGENSON, M. 1988. Rheology of melts and magmatic suspensions. 1. Design and calibration of concentric cylinder viscometer with application to rhyolitic magma. Journal of Geophysical Research, B93, 10273-10294. TANAKA, H. 2000. Viscoelastic phase separation. Journal of Physics: Condensed Matter, 12, R207-R264.

THOM, R. 1990. Apologie du Logos. Hachette, Paris. TUTUNCU, A.N., PODIO, A.L. & SHARMA, M.M. 1998. Nonlinear viscoelastic behavior of sedimentary rocks, Part II: Hysteresis effects and influence of type of fluid on elastic moduli. Geophysics, 63, 195-203. VANDERHAEGHE, O. & TEYSSIER, C. 2001. Partial melting and flow of orogens. Tectonophysics, 342,451-472. VIGNERESSE, J.L. & BURG, J.P. 2000. Continuous versus discontinuous melt segregation in migmatites: insights from a cellular automaton model. Terra Nova, 12, 188-192. VIGNERESSE, J.L. & TIKOFF, B. 1999. Strain partitioning during partial melting and crystallising felsic magmas. Tectonophysics, 312, 117-132. VIGNERESSE, J.L., BARBEY, P. & CUNEY, M. 1996. Rheological transitions during partial melting and crystallisation with application to felsic magma segregation and transfer. Journal of Petrology, 37, 1579-1600. WASSERMANN, G.S., ARNOL'D, V.I. & THOMAS, R.K. 1992. Catastrophe Theory. Springer-Verlag, Heidelberg. WEBB, S.L. & DINGWELL, D.B. 1990. Non-Newtonian rheology of igneous melts at high stresses and strain rates: experimental results for rhyolite, andesite, basalt and nephelinite. Journal of Geophysical Research, B95, 15695-15701. WILKS, K.R. & CARTER, N.L. 1990. Rheology of some continental lower crustal rocks. Tectonophysics, 182, 57-77. ZEEMAN, E.G. 1976. The umbilic bracelet and the double-cusp catastrophe. Structural Stability, the Theory of Catastrophes and Applications in the Sciences. Lecture Notes in Mathematics, 525, Springer-Verlag, Heidelberg, 328-366.

Index Page numbers in italics, e.g. 59, refer to figures. Page numbers in bold, e.g. 169, signify entries in tables. absolute plate motion (APM) compared to relative plate motion (RPM) 58-60, 59 model 29-30, 29, 30 Acapulco 118, 119 Acatlan 779, 722 Agua Blanca fault (ABF) 67, 172 Aguilas 250 Alboran Sea 250 Alicante 250 Almeria 250 Alpine fault 96, 199, 200, 209 Amami Sub-basin 300 Amami-Kagoshima Tectonic Line (AKTL) 299 Ammassalik erogenic belt (Am) 233 Animal Basin 67 Anita shear zone 200, 208 Anorituup Kangerlua 233 Arguello fault zone 67 Arrollo Taibena Basin 255 Arsuk 233, 235 Arsuk Brae 235 Arsuk Fjord 233, 235 Arthur River complex (ARC) 200 Asemi, River 287, 283, 285, 287 Aso, Mount 282 caldera 299 Atenango 722 attachment see coupling of limospheric layers Avaqqat Kangerluat 233, 237 Awatere fault 96 Balsas Basin 722 basal traction driven rotation 69 Baza 250 Beartooth Mountains (BT) 779 Beppu Bay 299, 305 Besshi nappe 287, 283, 285, 287 Betic Cordillera, Internal-External boundary hanging wall deformation 249-250, 273-275 evolution of structures and implications for coupling and decoupling 273, 274 geological map 250, 255-256, 260 geological setting 250-253 External Zones cross-section 252 pattern of vertical-axis rotations 272-273 regional constraints 271-272 rock successions 253-259 stratigraphy 254 structures 259-271 cross-sections 257-258, 261, 270 poles of bedding 265

Rambla Seca Basin 262 striations and fault planes 264 Bitterroot extensional complex (BC) 184, 186-187 Borgs Havn granite 235 brittle-ductile arc-parallel extension 279, 293-294 brittle deformation in Sambagawa 288 extreme ductile layer, normal thinning and arc-parallel stretching 288-290 exhumation scenario for Sambagawa 292-293, 293 normal fault development 289, 290 spacial distribution of recrystallized quartz grain sizes 291 strike-slip displacements 290 uniform shear sense and reversal by late-stage folding and faulting 290-291 variable strain geometry of exhuming rocks 292 five possible exhumation mechanisms 280 Butsuzo Tectonic Line (BTL) 282, 299 Cadiz 250 California see also San Andreas fault absolute plate motion (APM) and relative plate motion (RPM) 50, 59 Continental Borderland 65-66, 79-80 bending fault termination 76-78, 76 Pacific-North America tranform plate boundary 66-68, 67, 68 straight fault termination 69-76, 70 strike-slip fault termination styles 66, 68-78 vertical coupling and decoupling along WTR boundary 78-79 seismic anisotropy applicability of models 36-37 comparison between north and south California 35-36 model results 36 northern California 35 shear wave splitting measurements 70 southern California 32-34 shear-wave splitting 50 California, Gulf of 67 Camp Oven Creek (CO) 200 Capas Rojas Formation 266 Cartagena 250 Caswell Sound 799, 200, 209 paragneiss 208 P-T data 201 Cauta 250 Charles Sound 799, 200, 209

INDEX

338

Cheviot Hills 74 Chichibu metamorphic belt 281, 282 Chilapa 722 Chilpancingo 118, 119, 122 Chilpancingo Basin 722 Chino Hills 68, 75 Chugoku 143 Clarence fault 96 clutch tectonics 41-42, 51-52, 60 bottom-driven systems 53 implications 53 convergence 54, 57 divergence 54, 56-57 strike-slip partitioning and homogeneous mantle deformation 58, 58 transcurrent boundaries 53-56, 54 relation between crustal and mantle deformation 52 top-driven systems 52-53 Coacoyula 722 Coast Mountains-Cascades (CMC) batholith belt 169,

170 Coast shear zone (CSZ) 186-187 Cocula 722 Coeur d'Alene 184 Collnet Basin 67 Columbia River embayment (CRE) 779, 186-187 continental crust 313-314, 323 melting 314 melt segregation mechanisms and scales 314-315 P-T paths 315 near-isothermal decompression paths and mechanisms 315—316, 377 buoyancy/diapirism 317, 319 crustal thinning/collapse 316, 377 erosion 316, 377 exhumation of ultrahigh-pressure rocks 377, 318-319,378 folding/buckling 316-318, 377 low-angle normal faults 316 partial melting in orogens 319 buoyant return of subducted continental crust 322-323 isothermal decompression and migmatite domes 320-322, 327, 322 migmatite diapirs and gneiss domes 319-320 continental tectonics 1 Copalillo 722 Copalillo Basin 722 Cordoba 779, 250 coupling of lithospheric layers 1, 2 attachment formation during partitioning 231-232, 246 development of attachment zone 245-246 Psammite and Pelite Zones 244, 244 analogue modelling model construction 124-126, 725, 126

model results 129-132, 730, 737, 733, 734 model rheological structure and analogue materials 126-127,727 scaling of models 127-129, 128 qualitative comparison of model results with geology 134-136,735 model limitations 132-134 vertical coupling and decoupling 136 Cuautla 779 Cucamonga fault zone 68 Cuernavaca 118, 119 Cuevas del Ambrosio 268-269 Danell Fjord 233, 237, 239 Daniel, Mount (MD) 200 P-Tdata 201, 202 decollements 1-2 displacement transfer 177-178, 191 kinematic model 189-190, 189 decoupling of lithospheric layers 1, 2 diapirism 377, 319 dip-slip fault systems 1 Doubtful Sound 799 Dozan, River 287, 283, 285, 287 Eastern Gabar Basin 255 Eastern Transverse Ranges, California 92-94, 93, 94 Edgar, Mount (ME) 200 edge driven rotation 69 Egger 233 Elsinore fault 67 Elysian Park 74 Embalse de Valdeinfierno 262 fast direction polarization 9 Ferrelo fault 67 finite strain-controlled anisotropy 15-16 Fiordland, crustal attachment zone evolution 197-198, 223-226, 225 crustal structure and geochronology 207 age of magmatism, crustal melting and high-grade metamorphism 210 boundaries of high-grade metamorphic belt 207, 208, 209 geochronological data 203-206 lithological divisions of magmatic arc 207-210 structural relationships 210-212 evolution stage 1 - mafic-intermediate magmatism and partial melting of lower crust 213-215,274 evolution stage 2 - melt segregation and transfer mechanisms 215-218 (a) melt-induced fracture propagation 215-218, 277

(b) melt accumulation in ductile shear zones 218, 279 evolution stage 3 - evolving styles of deformation following magmatism and crustal melting 218

INDEX (a) steeply dipping sinistral and dextral shear zones 218-219 (b) gently dipping, layer-parallel shear zones 219, 227

(c) steep Indecision Creek and George Sound shear zones 220-223, 222, 223 geological map 200 geological setting 198-207 location map 799 P-T data 201-202 space-time correlation of high-grade fabrics 212-213 Foreland batholith belt 168 Fraser-Straight Creek fault (FSC) 186-187 Gabar 255-256 Garlockfault777 George Sound 799, 200, 209 P-T data 201 steep shear zone 220-223 Geurrero Morelos Platform 722, 123 Gibraltar 250 Goto Islands 299 Goto Sub-basin 299 Graenseland 233, 235 Granada 250 Guadalquivir Basin 250 Guadalupe fault zone 67 Guadalupe Microplate 67 Guadalupe Rift 67 Guadix 250 Hikimi 746, 750, 757 fault-slip data 155 Hiroshima City 148, 150 Hitoyoshi Basin 299, 305, 306 Hokkaido 143 Honshu 143, 299 Hope fault 96 Hosgri fault 67 Huajuapan 779 Huiziltepec 722 Igutsaat Fjord 233, 237 Ikermit 235 Indecision Creek 799 steep shear zone 220-223, 222, 223 Intermontane batholith belt 168 Ippatit 233 Ippatit Valley 239 Itoigawa-Shizuoka Tectonic Line (ISTL) 143 Izu Peninsula 143 Izu-Bonin Arc 143 Japan see also Sambagawa block rotation and intracrustal vertical decoupling 141-142, 158, 160

339

fault kinematics 152 age of recent inversion of motion sense 155-156 fault-slip data 754, 755 Plio-Quaternary to present-day kinematics 152-153 pre-Plio-Quaternary kinematics 154, 756 general geology 142-145 tectonic framework 143 SW island arc 297, 310 diffuse extension across south Kyushu 301-307 geodynamical and geological outline 297-301, 299 Okinawa-Kyushu junction area evolution model 307-310 Western Chugoku fault system 145-146, 148 age 152 analogue model similarities 156-158, 757 earthquake focal spheres 753 field occurrence 150-151, 757 formation model 159-160, 759 geometry of fault system 147-150 Kake-Himini area 750, 757 northern boundary 147, 146-147 Yamaguchi area 749, 757 Yoshiwa-Kake area 146 Julianehab batholith 232-234, 233 Border Zone-Julianehab batholith boundary 236, 237 construction and deformation 244 east coast structure 236-238 evolution 243 Julianehab batholith-Psammite Zone boundary 238 west coast structure 236 Kake 746, 750, 757 fault-slip data 755 Kamio, River 257, 283, 255, 287 Kangerluaraq 233, 237 Kangerluk 233, 237 Kangerluluk 233, 237 Kanoya Plain 299 Kap Farvel 233, 237 Kap Ivar Huitfeldt 233, 237 Ketilidian orogen, oblique convergence and attachment formation 231-232, 246 Border Zone structure Border Zone-foreland boundary relationships 234, 235 history 234-236 development of attachment zone 245-246 Julianehab batholith structure Border Zone-Julianehab batholith boundary 236, 237 east coast structure 236-238 west coast structure 236 major components 232, 233 Border Zone 232 Julianehab batholith 232-234

340

INDEX

Psammite and Pelite Zones 234 Psammite and Pelite Zones Julianehab batholith- Psammite Zone boundary 238 nature and timing of structural and metamorphic events 240 rapakivi suite 240-241 structure 238-240, 239 tectonic evolution 241, 242 construction and deformation of Julianehab batholith 244 development of mid-crustal attachment structure in Psammite and Pelite Zones 244, 244 rapakivi granite intrusion 245 structural histories 241, 243 Kettle extensional complex (KC) 779 Kikai Caldera 299 Kobberminebugt 233, 235, 243 Koshiki Islands 299 Kumamoto 252 Kyushu 143, 299 diffuse extension across southern regions 301, 305-307 Beppu region 305 Hitoyoshi-Ichifusa region 305, 306 Osumi region 301-305, 302, 303, 304, 305 geological structure 301 Okinawa-Kyushu junction area evolution model accomodation of extension through reactivation of thrust faults 307 cross-section model 307-310, 309 perpendicular extention 308 transtension in northern region 307 Laramide shortening, vertical coupling controlled by crustal heterogeneity 117-121, 137 lattice prefered orientation (LPO), olivine 3, 15 deformation types 55 mantle fabric observations 42-43, 44 Lewis and Clark lineament (LCL) 779, 184, 186-187 Lindenow Fjord 233, 237, 239 lithosphere, idealized deformation diagrams 54 Los Angeles 68 faults and seismicity 75 Los Angeles Basin 74 Malaga 250 Manapouri, Lake 799 mantle-driven deformation of orogenic zones 41-42, 60 crustal deformation and mantle fabric ancient orogens 48 interpretation of data 49-51 neotectonic orogens 49 strain history 51 lithosphere/asthenosphere connections continental settings 47-48 cratons 46

oceanic settings 46-47, 47 lithospheric deformation 42 mantle fabric laboratory experiments 43 numerical experiments 43-45 olivine LPO and shear-wave splitting 42-43, 44 mantle viscosity and seismic attenuation 45-46 Marbella 250 Maria 255-256, 260 Marlborough fault system, South Island, New Zealand 94-95, 96 Masuda City 148 Median Tectonic Line (MIL) 142,143,148, 282, 285, 299 Mexico, Caribbean-North American transform boundary 117-120, 137 analogue modelling of Late Cretaceous to Early Tertiary deformation model construction 124-126, 725, 726 model results 129-132, 130, 131, 133, 134 model rheological structure and analogue materials 126-127, 727 scaling of models 127-129, 128 geological and tectonic setting crustal structure 120, 727 Early Tertiary deformation 121-123, 722, 723 Laramide deformation 120-121 Tertiary deformation 124 lithological units 779 qualitative comparison of model results with geology 134-136, 135 model limitations 132-134 vertical coupling and decoupling 136 terrane boundaries 118 Midternass 233, 235 Milford 200 Milford Sound 799, 200, 209 P-T data 201, 202 Missoula 184 Mixteco Terrane 722 Mixteco-Oaxaca-Juarez block (MOJB) 120, 727 Early Tertiary deformation 121-123, 122 Laramide deformation 120-121 Tertiary deformation 124 Mogens Heinesen Fjord 233, 235 Montana disturbed belt (MDB) 779 Monterey fault zone 67 Monterey Microplate 67 Morro fault zone 67 mountain belts 1 Murcia 250 Nagssugtoqidian orogenic belt (Nag) 233 Nankai Trough 143, 299 Nanortalik 233 Napasorsuaq Fjord 233, 235, 237 New Zealand

INDEX absolute plate motion (APM) and relative plate motion (RPM) 50 seismic anisotropy 9-13, 31-32, 32-34 applicability of models 36-37 model results 36 modelling results 16-31, 17, 18-19, 20-21, 22

shear wave splitting measurements 10 study methods 13-16, 13 shear-wave splitting 50 Newport-Inglewood fault zone 68, 74-76 Niaqornaarsuk 233 Nobeoka Tectonic Line (NIL) 299 N0rrearm 233, 239 North American Cordillera 167-168, 168 crustal architecture 178-183 cross-sections 181-182 crustal thickening and deep flow 173 -174 crustal thickening and unroofing 168-169 Central Cordillera 169-170 Northern Cordillera 169, 170 Southern Cordillera 170-171 displacement transfer in decollement systems 177-178, 185-187, 191 generalized tectonic map 179 Late Cretaceous-Tertiary structural elements 186-187 oblique ramp system in Idaho-Montana basement 183-185, 184 coupling v. decoupling 190-191 influence on strike-slip systems 188 influence on Tertiary extensional systems 188-189 kinematic model for linked decollement system 189-190,759 regional expression 187-188 plateaux 171-172 unroofing mechanisms 172-173 Nunnarsuit 233 Oaxaca 118, 119 Oboke nappe 281, 283, 285, 287 Oita 282 Oita-Kumamoto Tectonic Line (OKTL) 252 Okanagan extensional complex (OC) 779 Okinawa trough back-arc basin 297, 310 diffuse extension across south Kyushu 301, 305-307 Beppu region 305 Hitoyoshi-Ichifusa region 305, 306 Osumi region 301-305, 302, 303, 304, 305 geodynamical and geological outline cross-section 300 Kyushi geological structure 301, 302, 303, 304, 305 present-day plate configuration and recent evolution 297-298, 299 structure 298-301

341

Okinawa-Kyushu junction area evolution model accomodation of extension through reactivation of thrust faults 307 cross-section model 307-310, 309 perpendicular extention 308 transtension in northern region 307 Olinala 722 olivine lattice preffered orientation (LPO) see lattice preffered orientation (LPO), olivine Omineca batholith belt 168 Omineca-Sevier batholith belt 168 Orizaba 118, 119 Orofino shear zone (OSZ) 779, 184, 186-187 orogenic float 1 erogenic zones, mantle-driven deformation 41-42, 60 crustal deformation and mantle fabric ancient orogens 48 interpretation of data 49-51 neotectonic orogens 49 strain history 51 lithosphere/asthenosphere connections continental settings 47-48 cratons 46 oceanic settings 46-47, 47 litho spheric deformation 42 mantle fabric laboratory experiments 43 numerical experiments 43-45 olivine LPO and shear-wave splitting 42-43, 44 mantle viscosity and seismic attenuation 45-46 Osburn fault 184 Oshima 282 Osumi Peninsula 299, 301-305 ages of pseudotachylite veins 303, 304, 305 cross-section of Osumi pluton 302 geological map 302 Otte Rud 0er 235 Oxnard 77 Oztotitlan Basin 722 Paatusoq 233, 237 Palos Verdes fault zone 68, 76-77 Papalutla 775 Papalutla Thrust 722 Patton Escarpment 67 Patton Ridge 67 Pembroke Valley (P) 200 P-T data 201, 202 Peninsular Ranges batholith (PRB) batholith belt 169, 772

Pinchi fault (PI) 186-187 Poison Bay 200 P-T data 201, 202 Priest River extensional complex (PC) 779, 184 Prins Christian Sund 233, 237 Puebla 778, 779 Puente Hills 75

342

INDEX

Puerto Angel 118, 119 Puisortoq 233, 235 Puisortoq Fjord 235 Qaqortoq 233 Qernertoq 233, 237 Qomoq augen granite 235 Qoornoq 235 Rambla Seca Basin 255, 262 cross-section 263 Raymond fault 68, 75 relative plate motion (RPM), compared to absolute plate motion (APM) 58-60, 59 rheology of partially molten rocks, strain-rate dependency 327-328, 334 basic rheological laws 328-329, 329 bulk response to low strain rates 332 bulk response to tectonic stress 331-332, 332 melt segregation at outcrop scale 333, 333 melt segregation within a vein 333-334, 333 pseudo-fluids 329-330, 330 two-phase materials 330-331 viscosity determination 334 Ryoke metamorphic belt 257, 282 Ryukyu Arc 299 Sakamoto antiform 281 Salina Cruz 779 Salton Trough 67 Sambagawa, flow patterns during exhumation of metamorphic rocks 279, 293-294 extreme ductile layer, normal thinning and arc-parallel stretching 288-290 exhumation scenario 292-293, 293 normal fault development 289 normal fault stereographs 290 spacial distribution of recrystallized quartz grain sizes 297 strike-slip displacements 290 uniform shear sense and reversal by late-stage folding and faulting 290-291 variable strain geometry of exhuming rocks 292 five possible exhumation mechanisms 280 geological outline 280-282, 287 geological map 282 study results 3D strain geometries 284-288, 284-285, 286 brittle deformation 288 mesoscopic structures 282-284, 283 shear sense distribution 287 San Andreas fault (SAF) 67 absolute plate motion (APM) and relative plate motion (RPM) 59 seismic anisotropy 9-13 Absolute Plate Motion (APM) model 29-30, 29, 30 modelling results 16-31, 77, 78-79, 20-21, 22

Pacific plate viscosity 29 shear wave splitting measurements 10 study methods 13-16, 73 symmetric weak fault model 23-25, 24-26, 24 San Benito fault 67 San Clemente fault 67, 70-73 aeromagnetic anomaly map 77 faults and seismicity in the Santa Barbara Channel area 73 seismic reflection profile 72 San Clemente Island 77 San Diego 68 San Fernando fault zone 75 San Gabriel fault 67, 68, 75 San Gregorio fault 67 San Isidro fault 67 San Jacinto fault 67 San Nicolas Island 77 San Pedro Basin 74 San Pedro Basin fault zone 73-74, 74 San Pedro Bay 74 San Quentin Basin 67 Santa Barbara 68 Santa Barbara Basin 77 Santa Barbara Channel 72-73 faults and seismicity 73 Santa Barbara fault 77 Santa Barbara Island 77 Santa Catalina Island 77 Santa Cruz Basin 67 Santa Cruz Island 77, 73 Santa Cruz-Catalina Ridge 77, 73 Santa Lucia fault 67 Santa Monica Basin 77, 74 Santa Monica Bay 74 Santa Monica-Hollywood fault zone 68 Sardlog shear zone 233 Saruta, River 287, 283, 285, 287 Satukujoq granite 235 seismic anisotropy New Zealand and California 9-13, 31-37, 32-34 applicability of models 36-37 comparison between northern and southern California 35-36 compression 34-35 modelling results 16-31, 77, 78-79, 20-21, 22, 36 study methods 13-16, 73 two-layer anisotropy 12 seismic attenuation 45-46 Sermilagaarsuk 233 Serreta de Guadalupe 255-256, 262 Seto Inland Sea 144, 148, 149, 281 Sevier batholith belt 168 Sevilla 250 shear-wave splitting 3 as a function of wave polarization 23

INDEX California 10, 50 changing depth to isotropy/anisotropy boundary 14 crustal block rotation by mantle flow 90-91 evolution of parameters 22 mantle fabric observations 42-43, 44 New Zealand 10, 50 oceanic material 46-47, 47 San Andreas fault (SAF) 10 Tibet 50 Trinidad 50 Venezuela 50 Shikoku Island 143, 281, 299 Shikoku-western Honshu region, Japan 144 Shimanto metamorphic belt 281, 282 Shimanto Terrane 306 Shimbara Peninsula 299 Shimokawa, River 281, 283, 285, 287 Shiraga, Mount 281 Shuswap extensional complex (SC) 779, 186-187 Sierra Nevada 10 Sierra del Gigante 255-256 Sierra del Maimon 255-256 Sierra del Pericay 255-256, 262 Sierra Larga 255-256 Sierra Madre fault zone 68, 75 Sierra Nevada (SN) batholith belt 169, 777 Sierra San Pedro Martir (SSPM) 772 Sikhote-Alin fault system 143 SKS phases 9-10 Snake River plain (SRP) 779 Snow Peak 184 Solana Formation 260 S0ndre Igaliku 233 S0ndre Sermilik 233, 243 Sorte Nunatak 233 Southern Japan Sea fault zone (SJSFZ) 143 strain gradients 101, 112-114 attachment tectonics 101-103, 102 modelling 103-104, 103, 104 transpression and transtension attachments 107-108, 109-112, 770, 773 transpression attachments 108-109, 777 transtension attachments 109, 772 wrench attachments 104-105, 105-107, 706, 707, 708, 709, 770 foliation and lineation patterns 105 strain modelling 9-13 California applicability of models 36-37 comparison between north and south California 35-36 model results 36 northern California 35 southern California 32-34 New Zealand 31 -32, 32-34 applicability of models 36-37 model results 36 results 16

343

effect of viscosity structure 22-30 other flow models 30-31 relative plate motion with isoviscous model 16-22,77,78-79,20-21,22 study methods 13-14 changing depth to isotropy/anisotropy boundary 14 model parameters 13 relation between deformation parameters and anisotropy 15-16 strike-slip fault systems 1 applicability of models 36-37 block rotation and intracrustal vertical decoupling 141-142, 158, 160 California comparison between north and south California 35-36 northern California 35 southern California 32-34 coupling at boundaries 9-13 modelling results 16-31, 77, 78-79, 20-21, 22 study methods 13-16, 73 model results 36 New Zealand 31-32, 32-34 partitioning 58, 58 termination at convergence zones 65-66, 79-80 bending fault termination 76-78, 76 Pacific-North America tranform plate boundary 66-68, 67, 68 straight fault termination 69-76, 70 termination styles 66, 68-78 vertical coupling and decoupling along WTR boundary 78-79 Tanakura Tectonic Line (TTL) 143 Tanegashima Island 299 Tasermiut 233 tectonic processes, implications of clutch tectonics 53 convergence 54, 51 divergence 54, 56-57 strike-slip partitioning and homogeneous mantle deformation 58, 58 transcurrent boundaries 53-56, 54 Tehuacan 779 Tehuantepec 779 Tehuantepec, Gulf of 118 Tibet absolute plate motion (APM) and relative plate motion (RPM) 50 lithospheric deformation 51 shear-wave splitting 50 Tintina fault (TI) 786-787 Tixtla 722 Tokara Line 299 Tokara Ridge 299, 300 Tokara Sub-basin 299, 300 Toluca 778

344

INDEX

transpressional zones 2 transpressional/transtensional attachment zones, strain gradients 101, 107-108, 109-112, 110, 112-114,113 attachment tectonics 101 -103,702 modelling 103-104, 103, 104 transpression attachments 108 -109, 777 transtension attachments 109, 772 wrench attachments 104-105, 105-107, 106, 107,

108, 109, 110 foliation and lineation patterns 105 transtensional deformation 297, 310 Okinawa-Kyushu junction area evolution model accomodation of extension through reactivation of thrust faults 307 cross-section model 307-310, 309 northern region 307 perpendicular extention 308 Trinidad absolute plate motion (APM) and relative plate motion (RPM) 50, 59 shear-wave splitting 50 Tsuneyama synform 257, 283 Tsushima fault system (TFS) 143 Tuliman 722 Tunua 235 Tuzantian Basin 722 Velez Blanco 255-256, 260 Velez Rubio 255-256 Venezuela absolute plate motion (APM) and relative plate motion (RPM) 50 shear-wave splitting 50 Verdugo fault zone 75 vertical axis rotations 3-4, 83-85, 84, 97-98 assumptions and applicability of model 97 Betic Cordillera 249-250, 273-275 evolution of structures and implications for coupling and decoupling 273, 274 geological map 250, 255-256, 260 geological setting 250-253, 252 pattern of vertical-axis rotations 272-273 regional constraints 271-272 rock successions 253-259, 254 structures 257-255, 259-271, 267, 262, 264, 265, 270 rigid rotations application to attachment/detachment zones 92 boundary conditions and background 85-87, 55, 56,57 experimental apparatus and design 87-89, 55, 59 experimental results 89, 90, 91 mantle deformation in obliquely convergent environments 91-92 natural systems 89-92 shear-wave splitting 90-91

side-driven v. bottom-driven systems 95-97 upper crustal rotation coinciding with mantle deformation 92, 94, 95 Eastern Transverse Ranges, California 92-94, 93, 94 Marlborough fault system, South Island, New Zealand 94-95, 96 Western Transverse Ranges (WTR), California 69 vertical coupling in the lithosphere channel flow 5 geophysical constraints shear wave splitting 3 vertical axis rotations 3-4 island arcs and marginal basins 4-5 orogenic belts and exposed attachment zones 4 viscosity and strain modelling effect of viscosity structure 22-23 Absolute Plate Motion (APM) model 29-30, 29,

30

asymmetric viscosity models 25-28, 27-25 effect of increasing compression component 28-29, 29 symmetric weak fault model 23-25, 24-26, 24 isoviscous model 16-22, 77, 75-79, 20-21, 22 Vizcaino Peninsula 67 western Idaho shear zone (WISZ) 779, 186-187 Western Metamorphic Belt (WMB) 777 western Nevada shear zone (WNSZ) 186-187 Western Transverse Ranges (WTR), California 66-69, 67, 79-80 clockwise vertical-axis rotation mechanisms 69 strike-slip fault termination styles 68-69 bending fault termination 76-78, 76 straight fault termination 69-76, 70 vertical coupling and decoupling along boundary 78 basal shear-driven block rotation 79 edge-driven block rotation 78-79 Whittier-Elsinore fault zone 77-78 Wind River Mountains (WR) 779 wrench attachments 104-105 foliation and lineation patterns 105 strain gradients 105-107, 106, 107, 108, 109, 110 Xochipala 722 Yakushi antiform 257, 253 Yakushima Island 299 Yamaguchi City 745, 749, 757 Yangsan fault system (YFS) 743 Yanhuitlan 775 Yoshiwa 746 Zarcilla de Ramos 255-256 Zarcilla de Ramos Basin 255 Zihuatanejo 775 Zitlala 722 Zumpango 722