Clay mineral formation and fabric development in the DFDP-1B borehole, central Alpine Fault, New Zealand

Abstract

Clay minerals are increasingly recognised as important controls on the state and mechanical behaviour of fault systems in the upper crust. Samples retrieved by shallow drilling from two principal slip zones within the central Alpine Fault, South Island, New Zealand, offer an excellent opportunity to investigate clay formation and fluid–rock interaction in an active fault zone. Two shallow boreholes, DFDP-1A (100.6 m deep) and DFDP-1B (151.4 m) were drilled in Phase 1 of the Deep Fault Drilling Project (DFDP-1) in 2011. We provide a mineralogical and textural analysis of clays in fault gouge extracted from the Alpine Fault. Newly formed smectitic clays are observed solely in the narrow zones of fault gouge in drill core, indicating that localised mineral reactions are restricted to the fault zone. The weak preferred orientation of the clay minerals in the fault gouge indicates minimal strain-driven modification of rock fabrics. While limited in extent, our results support observations from surface outcrops and faults systems elsewhere regarding the key role of clays in fault zones and emphasise the need for future, deeper drilling into the Alpine Fault in order to understand correlative mineralogies and fabrics as a function of higher temperature and pressure conditions.

Keywords:

• Alpine Fault
• clay minerals
• DFDP-1
• fault gouge
• fault rocks
• fault zone structure
• fluid–rock interaction

Introduction

Clay minerals can play an important mechanical role in active faults, as they are frictionally weak and their presence affects permeability with consequences for shear strength and fluid pressure evolution (e.g. Wu et al. 1975; Saffer & Marone 2003; Collettini et al. 2009). Along the surface exposures of the Alpine Fault in the South Island of New Zealand, the significance of clay minerals has been previously documented at Gaunt Creek. For example, Warr & Cox (2001) investigated the proportions of clay minerals in mylonite-derived cataclasites, and specifically of swelling clays (smectite) in the fault gouge at Gaunt Creek, whereas Boulton et al. (2012) reported significant amounts of illite and chlorite minerals in surface outcrops of fault gouge derived from both hanging-wall mylonite and footwall fluvio-glacial gravel in Gaunt Creek and the Waikukupa River. A particularly high amount of smectite (57–60 wt%) was detected relative to the units above and below, especially in the fault gouge unit at the Gaunt Creek Scarp exposure. The study yielded clay mineral proportions and experimentally measured frictional behaviours for samples that had been subjected to subaerial exposure for different times.

Here we document mineralogical and textural arrangements of clay minerals in material retrieved from the principal slip zone and the closely surrounding area of the Alpine Fault during shallow scientific drilling. The main goals of the Deep Fault Drilling Project (DFDP; Fig. 1) are to sample and monitor the Alpine Fault at depth, to better understand rock deformation, seismogenesis and earthquake deformation (Townend et al. 2009). The rocks analysed from the drillcore confirm the significance of fluid–rock interaction and alteration within the fault zone, and first results suggested the presence of an alteration zone that overprints the fault core and damage zone (Sutherland et al. 2012). In relation to the goals of current studies, the DFDP project provides an opportunity to characterise clay mineralisation in a rapidly exhuming, transpressive fault system by investigating the mineralogy of slip surfaces at depth. Our study of core from the DFDP-1B drillhole describes mineralogical characteristics and the fabric development of newly sampled rocks associated with active faulting, and provides the basis for conclusions regarding the mechanisms of clay formation and fault behaviour.

Figure 1 A, Location map showing the Australia–Pacific plate boundary through New Zealand. B, Map showing the Alpine Fault and other major tectonic features in the South Island. The star marks the DFDP-1 drill site (coordinates 43°17′5′′S, 170°24′22′′E).

The Alpine Fault and the DFDP-1 boreholes

The Alpine Fault is a dextral-reverse fault that fails in large (M_{w ~8}) earthquakes every 200–400 years and last ruptured in 1717 AD. It is an oblique transform boundary between the Pacific Plate and the Australian Plate through the South Island (Fig. 1) and is responsible for major earthquakes in the region, although none in historic times (Sutherland et al. 2007; Berryman et al. 2012). A long-term average strike-slip rate of 27 ± 5 mm a^–1 is accommodated by earthquakes, producing c. 8 m horizontal displacements (Norris & Cooper 2001; Sutherland et al. 2007). Recent palaeoseismological research at Hokuri Creek, at the southern end of the central section of the Alpine Fault, yields an average recurrence interval of 330 years (Berryman et al. 2012), meaning that a substantial fraction of the average time between successive earthquakes has elapsed since the last Alpine Fault earthquake almost 300 years ago. Interseismic alteration and healing of the fault zone is therefore in a mature state such that today's structure is likely representative of a pre-rupture state of the fault.

Two boreholes DFDP-1A and DFDP-1B were completed in February 2011 to depths of 100.6 m and 151.4 m, respectively, at Gaunt Creek (coordinates 43°17′5′′S, 170°24′22′′E) to characterise drilling conditions and sampling from shallow drilling (Townend et al. 2009; Fig. 1). On the south bank of Gaunt Creek, the Alpine Fault rocks are exposed in an outcrop almost 700 m long and >100 m high (Cooper & Norris 1994; de Pascale & Langridge 2012). A zone of cataclasites marks the fault in the hanging wall. In the southeastern part, cataclasite overlies fluvial gravel composed of rounded schist clasts; further northwest, the thrust has ridden over angular mylonite debris. The cores extracted during DFDP-1 drilling contain mylonites with Pacific Plate Alpine Schist protoliths, cataclasites derived from Pacific Plate mylonites and from Australian Plate felsic igneous rocks, gneisses and metasedimentary rocks, as well as gouge material from two principal slip zones (Sutherland et al. 2012; Toy et al. 2012). In addition to the coring operations, DFDP-1 involved the collection of wireline geophysical data (Townend et al. 2013) and hydraulic measurements (Sutherland et al. 2012), and the construction of a shallow fault zone observatory that provides ongoing measurements of temperature and fluid pressure as well as seismic monitoring (Sutherland et al. 2012).

Methods

Lithological and structural characteristics of five samples from borehole DFDP-1B were investigated in detail. We focused on the characterisation of newly formed (authigenic) clay minerals that can grow due to fluid–rock interaction processes and occur together with metamorphic phyllosilicates. Each sample's mineralogical composition was determined by X-ray diffraction (XRD) analysis using a Rigaku Ultima IV X-ray diffractometer with CuKα radiation and a step size of 0.02° 2θ. Bulk rock and clay preparation followed the analytical methods described in Moore & Reynolds (1997) and previous work on similar materials in fault zones elsewhere (Schleicher et al. 2009). Powder samples were prepared to analyse the bulk mineralogy, whereas oriented samples (<2 µm size-fraction) were used to identify the detailed clay mineralogy. Measurements were performed under air-dried and ethylene-glycolated conditions. The latter treatment causes interlayer expansion of swelling clays, allowing the recognition of discrete smectite and mixed-layer phases.

The presence of newly grown clay minerals has shown to be a key factor in the behaviour of active fault systems (Schleicher et al. 2010; van der Pluijm 2011). To quantify the amount of newly grown (authigenic) 1M_d and pre-existing (metamorphic) 2M₁ illite/muscovite in fault gouge and surrounding material in the clay-sized fractions, focused XRD scans were made on powdered samples between 16 and 44° 2θ on a Scintag X-ray diffractometer. The proportions of the two polytypes (1M_d and 2M₁) of the natural samples were calculated by comparing physical clay-standard mixtures with their calculated XRD patterns created using the program WILDFIRE© (Grathoff & Moore 1996; Haines & van der Pluijm 2008). These XRD patterns are used to iteratively model the percentage of the 2M₁ and 1M_d polytypes in several size fractions, each with its own ratio of metamorphic illite/muscovite to authigenic illite.

For scanning electron microscope (SEM) characterisation, gold-coated rock chips were examined with a Hitachi S3200N SEM using secondary electron imaging and semi-quantitative energy- (X-ray) dispersive spectroscopy (EDS). Selected samples were analysed using a Nova field-emission-gun SEM, in both backscattered and secondary electron modes.

The preferred orientation of clay minerals in the rocks was analysed by X-ray texture goniometry (XTG) using an Enraf-Nonius CAD4 automated single-crystal diffractometer with a Mo radiation source and a customised X-ray pole figure stage (van der Pluijm et al. 1994). Small rock-chips were cut into 0.2-mm-thick sections and glued on a sample holder. After the target clay phases (illite and chlorite) have been identified, the detector is moved to the specific angle (2Θ angle) of the target mineral, and the location and intensity of the diffracted beam is automatically collected in c. 1300 different positions. The resulting pattern is plotted in an equal-area projection and contoured as multiples of random distribution (m.r.d.) following established procedures (van der Pluijm et al. 1994). Here, higher m.r.d. values reflect higher degrees of crystallographic preferred orientation that can be compared with a growing catalogue of similar measurements from a range of settings and depths (e.g. Haines et al. 2009; Day-Stirrat et al. 2010, 2011).

Results

Borehole DFDP-1B has been described in detail and categorised into seven litho-tectonic units by Toy et al. (2012). The rocks penetrate a hanging-wall sequence containing mylonite mixed with cataclasite zones that increase in density downwards, 20–30 m of cemented and hydrothermally altered cataclasites and two principal slip zones (PSZ) formed of ultracataclasite and fault gouge. The footwall sequence contains quartz- and feldspar-rich, extensively altered and variably cataclased rocks including a slip zone at c. 144 m depth. Clay-rich fault gouge material was encountered in two principal slip zones within the DFDP-1B borehole: PSZ1 clay-gouge (sample 59/1/0.12 at 128.3 m in Fig. 2B, Unit 5 of Toy et al. 2012) lies below cataclasites of presumed Pacific plate provenance (Units 3 and 4) and above Pacific and Australian Plate-derived cataclasites (Unit 6); and PSZ-2 clay gouge (sample 69/2/46–66 at c. 144m in Fig. 2B) lies between mixed Australian and Pacific Plate cataclasites (Unit 6) and above Australian Plate-derived dark-grey brecciated mylonitic gneiss (Unit 7). The gouge samples taken show similar composition and microstructures to those documented by Boulton et al. (2012) in a nearby surface outcrop, as well as in drillcore from DFDP-1A and DFDP-1B (Boulton et al. 2014) with incohesive fault gouge material, partly containing rounded black clasts of ultramylonite and angular crushed clasts of pale-green cataclasite.

Figure 2 A, Sample locations in core from run 59, section 1 and run 69, section 2. B, Photos and description of selected rocks. Note that only three samples are shown here, but four samples are analysed in total in core run 69/2. Picture width c. 15 cm. C, X-ray diffraction patterns of clay size fraction of the samples, showing the occurrence of smectite in two principal slip zones (PSZ). Dark grey lines: air-dried analysis; light grey lines: ethylene glycolated. D, Changes of peak width measured in FWHM (full width at half maximum) in glycolated state. Note the similar thicknesses of chlorite and illite/smectite and broad FWHM in the slip zones.

In comparison to the fault gouge material, the cataclasite above PSZ-2 (sample 69/2/17–22 in Fig. 2B) is greenish-grey and highly fractured and contains thin dark-grey vein fillings. An ultracataclasite rock (sample 69/2/40–41), situated immediately above the dark-grey gouge zone of PSZ-2, contains gouge-filled veins forming foliations in a coarser-grained matrix- to clast-supported foliated cataclasite. The dark-grey fault gouge zone shows a distinct contact with the ultracataclasite. The cataclasite below PSZ-2 (sample 69/2/76–79) contains c. 90% matrix with outsized clasts that grades downwards into c. 30% matrix protocataclasite, composed mainly of angular to subangular clasts of black ultramylonite and white quartz. The abundance of deformed veins and mineralised fissures provide evidence that fluid pressures were locally high enough to stimulate fracturing (Toy et al. 2012).

The bulk rock mineralogy determined by X-ray diffractometry shows a composition of quartz, phyllosilicates (muscovite, chlorite), calcite, zeolite and clay minerals in all samples analysed. The dominant clay mineral phases are illite and chlorite/kaolinite (Fig. 2C). The characteristic peak for 001 kaolinite at c. 0.7 nm overlaps with the 002 chlorite peak: for this reason we here refer to both minerals. Based on the peak shape, as well as the occurrence of a weak kaolinite 002-peak at 0.4 nm (not shown here), both minerals most likely occur in all samples (see also Biscaye 1964). Moreover, Boulton et al. (2012) detected kaolinite in surface outcrops at Gaunt Creek. The clay mineral smectite is found solely in the fault gouge, comprising the two slip zones (sample 59/1/0.12 at 128.3 m and sample 69/2/46–66 at c. 144 m). The diffraction patterns of the clay-size (<2 μm) fraction of air-dried fault gouge of both samples show a very broad peak between the illite and chlorite peaks. Treatment with ethylene glycol shifted the peaks from c. 1.3 to c. 1.7 nm, which is characteristic of hydrated, three-water-layer Na-montmorillonite (Moore & Reynolds 1997). Montmorillonte was also detected by Boulton et al. (2014) in the fault gouge at c. 128.3 m. Illite and chlorite did not shift towards higher values, as they do not incorporate water in the interlayer sheets. Variations in the peak-width of (001) chlorite and (001) illite and/or smectite (measured as full width at half maximum, FWHM; Table 1) from XRD patterns of the clay-size fractions are shown in Fig. 2D. Both chlorite and illite show similarly narrow peaks in host rock mylonite above the PSZ-2 fault gouge, with half-width broadening (FWHM) values of 0.19–0.20 for chlorite and 0.21–0.29 for illite (Table 1). These narrow peaks reflect highly crystalline minerals and a very low abundance or absence of smectite interlayers. The fault gouge of PSZ-2 shows broad illite and chlorite peaks, reflecting smaller particles, and the presence of smectite. The dark-grey brecciated mylonitic gneiss beneath the PSZ-2 shows similar narrow peaks in comparison to material above the PSZ with FWHM values of 0.156 for chlorite and 0.101 for illite.

Table 1 X-ray diffraction data of chlorite and illite, with distinct peak positions (001/002 d-value), intensities (without background), peak-widths (FWHM) and 2θ angles.

Sample (DFDP-1B; core run/section/cm)	Depths (m)	Rock	Mineral	d-value (d)		Intensity	FWHM (glycolated)	2θ
59/1/0.12	c. 128.3	Fault gouge	Chlorite	d(001)	n.a.	n.a.	n.a.	n.a.
				d(002)	n.a.	n.a.	n.a.	n.a.
PSZ-1			Illite	d(001)	10.13	221	0.37	8.72
				d(002)	4.99	196	0.192	17.76
			Smectite	d(001)	17.6	1059	0.8	5.7
69/2/17–22	c. 143.5	Cataclasite	Chlorite	d(001)	14.29	2352	0.194	6.18
				d(002)	7.11	7857	0.187	12.44
			Illite	d(001)	10.02	6219	0.212	8.82
				d(002)	4.99	1652	0.195	17.74
69/2/40–41	c. 143.7	Cataclasite	Chlorite	d(001)	14.2	1067	0.2	6.22
				d(002)	7.09	3623	0.181	12.48
			Illite	d(001)	9.98	2316	0.229	8.86
				d(002)	4.99	509	0.231	17.76
69/2/46–66	c. 144	Fault gouge	Chlorite	d(001)	14.04	1163	0.426	6.2
				d(002)	7.09	2618	0.21	12.48
PSZ-2			Illite	d(001)	10.05	2389	0.25	8.8
				d(002)	4.99	592	0.18	17.76
			Smectite	d(001)	17.6	740	n.a.	5.7
69/2/76–79	c. 144.3	Breccia	Chlorite	d(001)	14.06	743	0.156	6.28
				d(002)	7.06	2042	0.167	12.52
			Illite	d(001)	9.97	5898	0.101	8.9
				d(002)	4.98	967	0.124	17.82

Quantification of metamorphic (2M₁) illite/muscovite and authigenic (1M_d) illite polytypes in the clay size fraction (<2 µm) of the fault gouge and the surrounding cataclasites is illustrated in Fig. 3 on the basis of XRD patterns and WILDFIRE© analysis of the measured spectra. All samples contain a mixture of both 2M₁ and 1M_d polytypes, and the intensity of the 2M₁ specific peaks (metamorphic grains) decreases consistently with a decreasing 2M₁ content of the mixtures. None of the patterns contain detectable feldspar peaks, which can overlap the polytypes' specific peaks used for modelling. In the two principal slip zone gouges (PSZ-1 and PSZ-2), the amount of newly grown (authigenic) illite is much higher (80–95% 1M_d) than the amount of metamorphic illite/muscovite (5–20% 2M₁), whereas the surrounding cataclasites show a much higher amount of metamorphic illite/muscovite (60–70% 2M₁) than authigenic illite (30–40% 1M_d).

Figure 3 Matches of calculated WILDFIRE patterns (grey line) and XRD standard mixtures with polytype patterns for cataclasite and fault gouge from the Alpine Fault (black line), allowing quantification of metamorphic illite/muscovite and authigenic illite polytypes.

Electron microscopy images of cataclasites and the gouge material show a mixture of less-altered pseudo-hexagonal illite, and strongly altered, rounded illite minerals adjacent to larger chlorite and muscovite grains with variable amounts of K, Ca, Mg and Fe (Fig. 4A–D). In the Unit 5 principal slip zone gouges, tiny smectite grains with strongly altered rims are present (Fig. 4C–D). Locally, very small Ca-rich illite and smectite grains are observed to have precipitated epitaxially on a quartz-feldspar-rich mineral matrix (Fig. 4C). In the cataclasite above PSZ-2, at 143.5 m, partially altered pseudo-hexagonal illite occurs in between feldspar, quartz and large chlorite phyllosilicates (Fig. 4A). Similar illite is found in the dark-grey brecciated mylonitic gneiss beneath PSZ-2; here, however, it also fills pores (Fig. 4B).

Figure 4 Secondary electron images showing clay minerals in two principal slip zones (PSZ) of Unit 5, the hanging wall and the footwall. A, Hanging-wall illite forming in a matrix of feldspar, quartz and chlorite (sample 69/2/40–41). B, Footwall illite growth as pore fillings (sample 69/2/76–79). C, PSZ-1 Ca-rich clay (smectite) growing on a feldspar surface. D, PSZ-2 smectite minerals occurring within a mixture of illite and chlorite.

Clay fabric intensity based on X-ray texture goniometry of both illite and chlorite in rock samples reveal that relatively strong fabrics are present in the cataclasites above the principal slip zone PSZ-2, yielding m.r.d. values of 3.3–3.7 (Fig. 5). Fabric intensities increase only slightly from the footwall towards the principal slip zone. The different clay mineralogy of the gouge imparts little effect on the measured fabric intensity; illite and chlorite in the gouge have very similar fabric intensities. In contrast, illite and chlorite in the PSZ-2 fault gouge show much lower fabric intensities (m.r.d. 2.2–2.7), typical of gouge material observed in other fault systems (see Haines et al. 2009). In the upper principal slip zone PSZ-1, the preferred orientation of illite and chlorite is so low (m.r.d. < 2) that fabric intensity could not be reliably quantified.

Figure 5 X-ray texture goniometry of samples from two PSZs and cataclasites, showing higher fabric intensities of illite in the cataclasite (multiple of random distribution m.r.d. = 3.3–3.7) than in slip zones (m.r.d. < 2.3).

Discussion

Fault gouge composition and clay formation

Low-temperature solution transfer during and/or after transient conditions of fault movement often leads to the alteration of the faulted rock and the formation of clay minerals (Wu et al. 1975; Evans & Chester 1995; Morrow et al. 2001; Kitagawa et al. 2007; Schleicher et al. 2009). The mineralogic composition of fault gouge is therefore a strong indication of the conditions that prevailed during fault rock formation. The unique occurrence of smectite mineralisation in the fault gouge of the two principal slip zones in the DFDP-1B drillhole indicates that recent fluids infiltrated along these zones at shallow depth during or after slip (see also Masuda et al. 1996; Schleicher et al. 2009). The localised concentration of smectite has received considerable attention in recent years as it affects frictional strength, resulting in weak fault behaviour (Ikari et al. 2007; Carpenter et al. 2011; Lockner et al. 2011). Similar smectite clay minerals were detected in other continental transform zones including, for example, the San Andreas Fault in Parkfield, California (Solum et al. 2006; Schleicher et al. 2006; Schleicher et al. 2010), where they can form a network of clay coatings on rock surfaces due to local infiltration of fluids during creep. In addition, Wu et al. (1975) observed smectite in fault gouge at the northern and central section of the San Andreas Fault Zone and along the Calaveras and Hayward faults, while Chester et al. (1993) showed that cataclastic and fluid-assisted processes in the cores of the Punchbowl and the San Gabriel faults are due to pervasive syntectonic alteration of the host rock minerals to zeolites and clays.

Other active faults, such as the Chelungpu Thrust Fault in Taiwan, show similar mineralogies as the Alpine Fault or the San Andreas Fault with smectite, illite and chlorite being the main clay minerals (Hashimoto et al. 2007; Isaacs et al. 2007; Sone et al. 2007). In this fault however, smectite was not found within the ultracataclastic deformed shear zone and in other gouge samples of the northern and the southern sites, but is abundant in the surrounding host rock (Boullier et al. 2009). Hashimoto et al. (2007) suggested that smectite consumption occurs due to dehydration and transition to mixed-layered illite-smectite during shear heating within the fault zone, with temperature the main factor influencing this reaction. Another possibility is that faulting itself supplies the energy required to overcome the kinetic barrier and transform smectite into illite (Vrolijk & van der Pluijm 1999; Isaacs et al. 2007). In both cases, the relation of smectite consumption is most likely related to fault activity.

The unique presence of smectite in the principal slip zones in the Alpine Fault drillcore indicates a low-temperature environment during its formation, as this mineral is not stable at temperatures much above c. 150 ° C (e.g. Abercrombie et al. 1994). Additionally, based on the large proportion of authigenically grown 1M_d illite together with distinct alteration features recorded in SEM images, these minerals were likely formed by precipitation reactions during movement of fluids along localised fractures, at the expense of host-rock minerals. Clay minerals in cataclasites and ultracataclasites above and below the DFDP-1B principal slip zones (Unit 3, 4, 6 and 7 of Toy et al. 2012; Townend et al. 2013) contain no smectite. Here, the host-rock alteration minerals formed as vein fillings or as alteration products from feldspar or mica (Fig. 2A–B). The marked differences in mineralogy suggest a different mechanism and conditions of formation.

The preferred orientation of phyllosilicates in natural faults has been argued to be a key control on fault frictional behaviour (e.g. Rice 1992). In the Alpine Fault, the fabric intensity in both slip zones is uniformly weaker than in the adjacent rock (m.r.d. 2.2 in the fault gouge compared to m.r.d. 3.7 in the cataclasite; Fig. 4). Both illite and chlorite also show similarly weak fabrics in the principal fault gouge of Unit 5, consistent with measurement of fault gouges elsewhere, with an average fabric intensity of 2.6 m.r.d. or less (Haines et al. 2009). The weak clay fabrics in Alpine Fault gouge support the notion that clay orientation in these fault rocks is a result of shallow authigenic mineral growth and not strain-induced particle reorientation of pre-existing phases. Weak fabrics also indicate that fluids were not focused along the fault by anisotropic permeability of rock fabrics, but more likely by variably spaced and interconnected fracture networks (e.g. Schleicher et al. 2010). Haines et al. (2009) suggests two possible explanations for the characteristically low fabric intensities in smectite-rich clay gouges. The very weak fabric could have formed as a result of clays that grew after fault slip ceased. As the fault surface was tectonically exhumed to a depth where differential stresses are lower, perhaps this led to a more isotropic clay fabric. Alternatively, the weak fabrics may represent a fundamental property of authigenic clay growth in active fault zones. Smectite consistently shows a distinctive wavy and irregular habit that is visible at very high magnifications, whereas illite and chlorite at the same magnifications typically have more planar morphologies. It is therefore likely that the low fabric intensities for smectite in Alpine Fault gouge are partly a function of the crystallite morphology of smectite and the variable orientation of precipitation surfaces. This hypothesis can be tested by additional study of deeper gouge that will be sampled in the next phases of DFDP.

Conclusions

Analysis of a small suite of samples collected in the first phase of scientific drilling into the Alpine Faults yields observations that support results from surface exposures and active fault zones elsewhere. Mineralogical data and microstructural observations provide evidence for extensive fluid–rock interaction along the Alpine Fault. Gouge from the principal slip zone of the Alpine Fault in the near-surface is smectite-rich and has little to no internal clay fabric. We observe that smectite formation is associated with transformation of very fine-grained material produced by cataclasis during slip, and the occurrence is restricted to a zone where slip occurs. We surmise that there may be a positive feedback between slip-weakening and smectite formation related to the low dynamic strength of smectite and/or dynamic pressurisation during slip.

Our analysis of samples retrieved by DFDP-1 drilling and sampling contributes to a growing body of evidence that chemical alteration, and smectitic clay mineral formation in particular, play a significant mechanical role within active shallow faults.

Acknowledgements

The study benefited from discussions with Brett Carpenter and Caroline Boulton and journal reviews by James Evans and an anonymous reviewer. This research was supported by NSF grant EAR1118704 (BvdP). DFDP-1 drilling and sampling (principal investigators RS, JT and VT) was funded by GNS Science; Victoria University of Wellington; the University of Otago; the University of Auckland; the University of Canterbury; Deutsche Forschungsgemeinschaft and the University of Bremen; Natural Environment Research Council and the University of Liverpool; and the Marsden Fund of the Royal Society of New Zealand.