Peak metamorphic temperature and thermal history of the Southern Alps (New Zealand)

HAL Id: hal-02133783

https://hal.archives-ouvertes.fr/hal-02133783

Submitted on 19 May 2019

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Peak metamorphic temperature and thermal history of the Southern Alps (New Zealand)

O. Beyssac, S.C. Cox, J. Vry, F. Herman

To cite this version:

O. Beyssac, S.C. Cox, J. Vry, F. Herman. Peak metamorphic temperature and thermal his- tory of the Southern Alps (New Zealand). Tectonophysics, Elsevier, 2016, 676, pp.229-249.

�10.1016/j.tecto.2015.12.024�. �hal-02133783�

Peak metamorphic temperature and thermal history of the

1

Southern Alps (New Zealand)

2 3

Beyssac O.

, Cox S.C.

, Vry J.

, Herman F.

4

5

(1) Institut de Minéralogie, de Physique des Matériaux, et de Cosmochimie, UMR 6

CNRS 7590, Sorbonne Universités – UPMC, Muséum National d’Histoire 7

Naturelle, IRD, 4 place Jussieu, 75005 Paris, France 8

(2) GNS Science, Private Bag 1930, Dunedin, New Zealand 9

(3) Victoria University of Wellington, P O Box 600, Wellington, New Zealand 10

(4) Institute of Earth Surface Dynamics, University of Lausanne, Switzerland 11

12

* Corresponding author: Olivier.Beyssac@upmc.fr 13

14

Submitted to Tectonophysics. Word count ca 14400 all included, 12 Figures, 2 tables.

15 16 17

Abstract

18

The Southern Alps of New Zealand result from late Cenozoic convergence between 19

the IndoAustralian and Pacific plates, and are one of the most active mountain belts in 20

the world. Metamorphic rocks carrying a polymetamorphic legacy, ranging from low- 21

greenschist to high-grade amphibolites, are exhumed in the hanging wall of the 22

Alpine Fault. On a regional scale, the metamorphic grade has previously been 23

described in terms of metamorphic zones and mineral isograds; application of 24

quantitative petrology being severely limited owing to unfavourable quartzo- 25

feldspathic lithologies. This study quantifies peak metamorphic temperatures (T) in a 26

300 x 20 km area, based on samples forming 13 transects along-strike from Haast in 27

the south to Hokitika in the north, using thermometry based on Raman spectroscopy 28

of carbonaceous material (RSCM). Peak metamorphic T decreases across each 29

transect from ≥ 640°C locally in the direct vicinity of the Alpine Fault to less than 30

330°C at the drainage divide 15-20 km southeast of the fault. Thermal field gradients 31

exhibit a degree of similarity from southernmost to northernmost transects, are greater 32

in low-grade semischist than high-grade schist, are affected by folding or 33

discontinuous juxtaposition of metamorphic zones, and contain limited information on 34

crustal-scale geothermal gradients. Temperatures derived by RSCM thermometry are 35

slightly (≤ 50°C) higher than those derived by traditional quantitative petrology using 36

garnet-biotite thermometry and THERMOCALC modeling. The age of RSCM T 37

appears to be mostly pre-Cenozoic over most of the area except in central Southern 38

Alps (Franz Josef-Fox area), where the amphibolite facies schists have T of likely 39

Cenozoic age. The RSCM T data place some constraints on the mode of exhumation 40

along the Alpine Fault and have implications for models of Southern Alps tectonics.

41 42

Keywords 43

44

Southern Alps; Alpine Fault; RSCM thermometry; Alpine Schist; exhumation

45

46

1. Introduction 47

48

The kinematics and thermal structure of orogenic wedges result from the coupling 49

between crustal and surface processes at convergent plate boundaries. Being one of 50

the most active mountain belts in terms of both tectonic and surface processes, the 51

Southern Alps of New Zealand offers a unique tectonophysical laboratory to 52

investigate these interactions. The rocks of this mountain belt were formed by 53

Paleozoic and Mesozoic subduction-accretion processes at the paleo-Pacific margin of 54

Gondwana, split from Gondwana and were thinned during the Late Cretaceous, then 55

rent by dextral strike-slip displacement as the Alpine Fault plate-boundary developed 56

during the Neogene.

57

The Southern Alps, which comprise much of the South Island (Figure 1), 58

began forming during the late Cenozoic as the IndoAustralian-Pacific plate motion 59

became increasingly convergent in the Pliocene-Pleistocene. These mountains form 60

against the Alpine Fault - a transpressive section of the Pacific and IndoAustralian 61

plate boundary (see Cox and Sutherland, 2007 for review). The Pacific Plate presently 62

appears to delaminate (e.g. Molnar et al., 1999) or subduct (e.g. Beaumont et al. 1994) 63

within the orogen, actively exhuming a belt of mid-upper crustal material obliquely 64

on the Alpine Fault, and accreting lower crustal material into a thickened crustal root 65

(e.g. Gerbault et al., 2002). The plate boundary is widely cited as a type-example of 66

deep geological processes and continent-continent collision (e.g. Okaya et al., 2007).

67

Over the past twenty years, there has been considerable scientific effort trying 68

to understand the architecture of the IndoAustralian-Pacific plate convergence in the 69

South Island (e.g. Okaya et al., 2007). Evidence has been gathered on the depth of the 70

crustal root, nature of lithosphere, and geometry of faults (see Okaya et al., 2007).

71

This effort has been complemented by thermochronologic work to decipher the timing 72

and thermal structure associated with mountain building and exhumation (e.g.; Tippett 73

and Kamp, 1993a,b; Batt et al., 2000; Herman et al., 2009). Other studies have noted 74

perturbations of the geotherm, producing high thermal gradients and hot spring 75

activity (e.g. Allis et al., 1979; Koons 1987; Allis and Shi 1995; Sutherland et al., 76

2012; Cox et al., 2015). However, while the general metamorphic structure of the 77

Southern Alps is qualitatively well established, there are very few quantitative 78

constraints on the thermal state and thermal history of the crust. An understanding of

79

the thermal history of the orogen is needed to constrain the information low- 80

temperature thermochronometers provide about erosion rates and the stability of 81

landforms, as well as the rheology of rocks, behavior of faults at seismogenic depth 82

(Toy et al., 2010), and ultimately seismic hazard (Sutherland et al., 2007). The lack of 83

thermal state information is largely attributable to the bulk rock compositions (mainly 84

metamorphosed quartzofeldspathic greywacke) that are chemically unfavourable for 85

precise metamorphic petrology, and complicated further by the polymetamorphic and 86

polydeformational history of the rocks and potential overprinting effects of fluid flow 87

(Koons et al., 1998; Vry et al., 2004; Menzies et al., 2014).

88

In this study we introduce thermometry based on Raman spectroscopy of 89

carbonaceous material (RSCM) (Beyssac et al. 2002) that allows the quantitative 90

estimate of peak metamorphic temperature (T) independently from the extent of 91

retrogression and presence of diagnostic mineral assemblages. Owing to widespread 92

presence of carbonaceous material in the local Alpine Schist and greywacke, this 93

technique has enabled the generation of a large dataset covering most of the Alpine 94

Fault hanging wall, both along strike and perpendicular to the fault. We present a 95

dataset of 142 new temperature estimates covering a 300 x 20 km area (Table 1). We 96

have also revisited traditional garnet-biotite thermometry results for some of the same 97

samples used for RSCM thermometry, or collected from nearby locations. We 98

provide those results for comparison, along with a few insights gained through 99

comparison of the observed mineral assemblages with their stability fields in P-T 100

pseudosections calculated using THERMOCALC. We then discuss the age of these 101

temperatures by reviewing existing geochronologic constraints to separate the 102

Mesozoic legacy from the late Cenozoic thermal overprint and the extent to which 103

this varies along the plate boundary. Finally, we highlight some constraints these 104

RSCM temperature distributions place on the style and nature of Southern Alps 105

tectonics.

106 107

2. Geological setting 108

109

2.1. General tectonics of the Southern Alps 110

Figure 1 depicts simplified geological and topographic maps of the South Island.

111

Pacific Plate motion relative to the IndoAustralian Plate is 39.7 ± 0.7 mm/a at 245 ±

112

1° in the central South Island (MORVEL model of DeMets et al., 2010). The vector is 113

12° anticlockwise of the Alpine Fault, which strikes 053° and is inferred to dip ~40- 114

60° SE (Norris and Cooper, 2007; Stern et al., 2007), extending downward to depths 115

of 25–30 km based on the presence of amphibolite facies schist exhumed in its 116

hanging wall (Grapes, 1995). The generally accepted crustal model depicts the Alpine 117

Fault shallowing eastward into a lower crustal décollement that delaminates the 118

Pacific Plate (Figure 2, e.g., Wellman, 1979; Norris et al., 1990; Okaya et al., 2007), 119

although there is no conclusive evidence for such a detachment. Thermochronological 120

modeling indicates uplift/cooling must be a two stage process first initiating on a 121

gently rising trajectory beneath the dry pro-side of the mountains then occurring 122

more-rapidly up the Alpine Fault ramp (Herman et al., 2009). While the maximum 123

metamorphic grade of exhumed rocks has been used to infer the approximate depth of 124

the Alpine Fault and Pacific Plate delamination, it is predicated on an assumed 125

geothermal gradient and the assumption that previously-stable metamorphic 126

assemblages were exhumed in the late Cenozoic. Although low-temperature 127

thermochronologic ages are clearly the result of Neogene-Quaternary cooling and 128

exhumation, many of the rocks reached peak metamorphic temperatures during the 129

Mesozoic, so much care is needed when using metamorphic assemblages to constrain 130

the present crustal structure.

131 132

2.2. Tectonostratigraphy and structural framework 133

Rocks of the Pacific Plate, southeast of the Alpine Fault, belong to the Eastern 134

Province and mostly to the Torlesse Composite Terrane/Supergroup (Mortimer et al., 135

2014). They are dominated by compositionally monotonous greywacke sandstone 136

and argillite sequences of the Rakaia Terrane that were deposited in an accretionary 137

prism on the margin of Gondwana during the Permian-Triassic (Mortimer, 2004).

138

Sediments were tectonically stacked and imbricated by accretion, and now locally 139

include some metavolcanic-rich and Cretaceous sequences (Cooper and Ireland, 140

2013). Metavolcanic, chert and micaceous-rich schist sequences have been 141

differentiated locally as the Aspiring lithologic association (Craw, 1984; Nathan et al., 142

2002; Cox and Barrell, 2007; Rattenbury et al., 2010) and may warrant separate 143

terrane status (Cooper and Ireland, 2013). Regional low-grade metamorphism affected 144

much of the Rakaia Terrane during the Jurassic (Mortimer 1993, 2000), possibly

145

involving several discrete metamorphic events (Adams 2003; Adams and Maas 2004).

146

Schist fabrics and most metamorphic mineral growth have occurred during Jurassic- 147

Cretaceous metamorphism (Cooper and Ireland, 2013; Mortimer, 2000; Vry et al., 148

2004).

149

The 300 km-long belt of schist adjacent to the Alpine Fault contains multiple 150

generations of metamorphic fabrics, folds, and syn- to post-metamorphic quartz veins 151

(Little et al. 2005; Cox and Barrell 2007). A near-continuous mid-upper crustal 152

section is exposed southeastwards across the Southern Alps (Grapes, 1995; Grapes 153

and Watanabe 1992; Little et al., 2005). Mid-crustal mylonites and amphibolite facies 154

schist adjacent to the fault contain evidence of a late Cenozoic ductile deformation 155

overprint that constructively reinforced and reoriented the pre-existing Mesozoic 156

metamorphic fabrics (Little et al. 2002a, 2007; Norris and Cooper 2003, 2007; Toy et 157

al., 2008). A steeply dipping array of late-stage shears that is present for 20 km in the 158

Franz Josef – Fox area (central Southern Alps) represents an exhumed, fossil, brittle- 159

ductile transition zone (BDTZ; Little et al., 2002b; Wightman and Little, 2007). The 160

zone separates schist from relatively undeformed greywacke and semischist 161

sequences that were metamorphosed during the Mesozoic but suffered only brittle 162

effects during the late Cenozoic IndoAustralian-Pacific plate transpression (Cox et al., 163

1997, 2012).

164

Two regionally extensive foliations are developed in schist beside the Alpine 165

Fault: An early foliation (S1 or S2) that is sub-parallel to remnant bedding and 166

metamorphic isograds, and a steeply dipping crenulation foliation (S3) that is axial 167

planar to folds of the S1/S2 fabric and bedding (Grindley, 1963; Little et al. 2002a).

168

At regional scale, metamorphic mineral zones are slightly oblique to boundaries in the 169

textural development of schistosity and cleavage (Little et al., 2005; Cox and Barrell, 170

2007; Rattenbury et al., 2010). The belt of schist varies in width along the orogen, 171

being narrowest (8 km) in the Franz Josef – Fox area of the central Southern Alps, 172

where tight to isoclinal folding has aligned the early schistosity with 035-045º strike 173

weakly oblique to the 055º Alpine Fault (Little et al., 2005). Elsewhere, the folding is 174

more open, early schistosity strikes at a high-angle (nearly perpendicular) to the 175

Alpine Fault, and the schist belt broadens. Mineral metamorphic grade changes are 176

indicated by the presence or absence of key minerals or mineral assemblages but 177

textural metamorphic zones, a semi-quantitative measure of the degree of cleavage, 178

foliation and metamorphic segregation development, are considerably easier to apply

179

and map in the field (Bishop, 1974; Turnbull et al., 2001). This classification is 180

widely used throughout New Zealand, and offers a first-order representation of the 181

increasing deformation gradient on a regional scale towards the Alpine Fault. From 182

SE to NE, following increasing metamorphic grade, there are four main textural 183

zones: uncleaved greywacke (TZ1), cleaved greywacke (TZ2a) to well cleaved semi- 184

schists (TZ2b), foliated schists (TZ3) and well segregated schists (TZ4).

185 186

2.3. Metamorphism and geochronology in the Alpine Schist 187

The general metamorphic pattern is well mapped and qualitatively constrained in the 188

Southern Alps by the presence or absence of key index minerals or mineral 189

assemblages (Nathan et al., 2002; Cox and Barrell, 2007; Rattenbury et al., 2010).

190

There is a general SE-NW increasing metamorphic gradient, ranging from sub- 191

greenschist (prehnite-pumpellyite or pumpellyite-actinolite facies); to chlorite then 192

biotite zones (greenschist facies); and garnet-oligoclase then K-feldspar zones 193

(amphibolite facies). Metamorphic mineral zones have been mapped on the basis of 194

the first appearance of minerals, but many of the ‘isograd’ boundaries between zones 195

represent juxtaposition by brittle faults or ductile shear zones, such that the ‘first 196

appearance’ of an index mineral cannot necessarily be assumed to represent a 197

preserved mineral reaction surface (Craw, 1998). Blocks with distinct metamorphic 198

and structural histories are commonly bound by faults or shear zones. Locally, distinct 199

phases of biotite and garnet crystal growth can be distinguished. Fine-grained biotite 200

and very fine-grained grossular-spessartine garnet are associated with rocks attaining 201

garnet-biotite-albite zone of the greenschist facies during the Jurassic, and are 202

distinguished by the biotite-1 and garnet-1 ‘isograds’ (White, 1996; Mortimer, 2000;

203

Cox and Barrell, 2007; Rattenbury et al., 2010). Whilst common in Otago, in most 204

places in the Southern Alps these minerals were either (i) completely overgrown or 205

consumed by growth of porphyroblastic ‘biotite-2’ and ‘garnet-2’ almandine as 206

metamorphism reached amphibolite facies during the Cretaceous-Cenozoic, or (ii) 207

have been retrogressively replaced by chlorite. As with textural zonation, the Alpine 208

Schist metamorphic mineral zones are slightly oblique to the Alpine Fault and to the 209

main SW-NE structural trend of the mountains.

210

There are few detailed and quantitative petrology studies on Alpine Schist. P–

211

T estimates have been made using garnet–biotite thermometers and barometers that 212

involve partitioning of Ca between garnet and plagioclase (Cooper, 1980; Grapes and

213

Watanabe, 1992; Grapes, 1995). Peak metamorphic conditions have been constrained 214

to 600–700 °C and 9.2–10 kbar in the Franz Josef – Fox area (Grapes and Watanabe, 215

1992; Grapes, 1995). Such high temperatures were recently confirmed by Toy et al.

216

(2010) applying Ti-in-Biotite thermometry in the mylonitic rocks from the Alpine 217

Fault shear zone. Within a few kilometers of the Alpine Fault, peak P-T conditions 218

decrease progressively to lower P–T conditions, e.g. 400–540 °C and 4–7 kbar 219

towards the southeast (Cooper, 1980; Green, 1982; Grapes, 1995; Grapes and 220

Watanabe, 1992). To the north, in the Hokitika region, Vry et al. (2008) established 221

one of the rare P-T paths available for Alpine Schist, using pseudosection 222

calculations. They obtained a prograde P-T path from ca. 380°C / 2.5 kbar to ca.

223

490°C / 8.5 kbar followed by a slight increase of temperature during decompression 224

to reach ca. 500°C / 6.5 kbar.

225

Geochronological and thermochronological datasets available for the Southern 226

Alps cover a range from low to high closure temperature systems. Batt et al. (2000) 227

generated a compilation of new and previously available fission track data from 228

zircon and/or apatite (see also Tippett and Kamp, 1993a,b, 1995), and K-Ar and

Ar- 229

39

Ar data from muscovite, biotite and/or K-feldspar. This compilation was 230

subsequently complemented in Herman et al. (2009). These studies illustrated 2- 231

dimensional patterns of thermochronological ages with variation both along-strike and 232

perpendicular to the Alpine Fault. Along strike, old apparent ages are almost 233

systematically observed with all techniques close to Haast River in the southwest, 234

representing a complex burial and exhumation history during the Cretaceous and 235

Cenozoic, prior to the development of the modern tectonic regime. Ages decrease 236

towards the northeast and are mostly late Cenozoic and ‘reset’ in Alpine Schist of the 237

central Southern Alps. For instance, Chamberlain et al. (1995) derived

Ar-

Ar dates 238

for hornblende in the Alpine Schist and obtained reset Neogene (3-6 Ma) ages in the 239

Fox-Franz Josef area, as well as some disturbed 25 Ma ages with plateau ages 240

suggestive of partial gas retention. Although two Cretaceous

Ar-

Ar ages on 241

hornblende were subsequently obtained by Little et al. (2005) in Fox Valley and 242

Waikukupa River, other hornblende ages and all other thermochronometers for schist 243

between Copland and Whataroa Rivers record a Neogene exhumation history for the 244

central Southern Alps. Further northeast, higher temperature thermochronometers 245

record increasingly older ages, partially reset by the thermal regime of the present 246

plate boundary. Perpendicular to the Alpine Fault, all thermochronometers show a

247

zone of young ages adjacent to the fault, which increase southeastwards across the 248

mountains through a series of exhumed partial annealing zones. The pattern of 249

resetting is systematic, with annealing zones located further from the Alpine Fault, but 250

also along strike away from the central Southern Alps, with decreasing closure 251

temperature of the thermochronometer (Batt et al. 2000; Herman et al., 2009; Little et 252

al., 2005).

253

The long and complex burial and exhumation history of Alpine Schist makes it 254

hard to decipher and distinguish the thermal history during late Cenozoic evolution of 255

the plate boundary, and/or formation of the Southern Alps (Mortimer, 2004; Cox and 256

Sutherland, 2007). Schists in the Southern Alps are generally thought to be correlated 257

with those in Otago – both being part of the Haast Schist Group (see Mortimer et al., 258

2014). But while metamorphic assemblages in Otago were formed during Jurassic- 259

Early Cretaceous and schist uplifted in the Early Cretaceous, some metamorphic 260

growth in Alpine Schist occurred during the Late Cretaceous and a component of 261

ductile-brittle deformation overprinted, but did not involve complete recrystallization 262

of these fabrics during the Neogene (Little et al., 2002a,b). A few studies have 263

suggested that locally in the central Southern Alps, peak metamorphic temperatures 264

may have been reached in the Alpine Schist during the Neogene. For instance, on a 265

large scale, by reconsidering available thermochonological data in light of geological 266

and geophysical observations, Little et al. (2005) proposed the local 'Alpine' 267

exhumation of lower crustal rocks (T>500°C) in a narrow zone where

Ar-

Ar ages 268

on hornblende are totally reset. This narrow zone is located between Franz Josef – 269

Fox and is not located right on the Alpine Fault but a few kilometers from it to the SE.

270

Combining detailed petrological and geochronological investigations on garnet 271

porphyroblasts collected farther to the north, from schist near the Alpine Fault in the 272

Hokitika region, Vry et al. (2004) were able to show that at least locally, peak 273

metamorphic temperatures of ca. 600°C were reached during the late Cenozoic and 274

recorded in the external growing zone of garnets. For this, they dated the different 275

zones of the garnet porphyroblasts using Sm-Nd and Lu-Hf systems and calculated for 276

each zone the P-T conditions of crystallization using garnet-biotite thermometry and 277

garnet-plagioclase-muscovite-biotite barometry.

278 279

3. Methods

280

281

3.1. Sampling 282

Samples collected for this study have been contributed to the national New Zealand 283

rock and mineral collection PETLAB (pet.gns.cri.nz). Samples already in the 284

collection, obtained during many years of mapping and research investigations in the 285

western Southern Alps (e.g. Grindley, 1963; Cox and Barrell, 2007; Rattenbury et al., 286

2010), were also analysed. Schist and semischist samples have been systematically 287

characterized for their lithology, main mineral assemblage, textural fabric 288

development, and are georeferenced with standard coordinate systems. Any research 289

action on the samples, such as geochronology, geochemistry or quantitative petrology, 290

has been recorded, and substantial information has been returned to the PETLAB 291

database where it is publically accessible. Supplementary material contains a kmz file 292

of samples, sample descriptions and RSCM T results. Numbers prefixed by P, OU or 293

VU refer to unique sample identifiers used by PETLAB.

294

Samples were collected from the field, or selected from PETLAB, to form 295

thirteen transects across the Southern Alps named according to major rivers or places, 296

to quantify variations both across and along the plate boundary. For each sample 297

location, we calculated a three-dimensional structural distance D (in km) to the Alpine 298

Fault plane by assuming the fault dips at 45° from its mapped surface trace and using 299

the topographic altitude of the sample site, as depicted on Figure 2. Possible variation 300

in D, associated with uncertainty in the subsurface geometry of the fault, was also 301

assessed by varying the dip by ±15°, to 30° and 60° (see below).

302 303

3.2. RSCM thermometry 304

RSCM thermometry is based on the quantitative study of the degree of graphitization 305

of carbonaceous material (CM) that is a reliable indicator of metamorphic temperature 306

(T). Because of the irreversible character of graphitization, CM structure is not 307

sensitive to the retrograde path during exhumation of rocks and records the maximum 308

T reached during metamorphism (Beyssac et al., 2002). Absolute T can be determined 309

in the range 330-650°C with a precision of ± 50 °C due to uncertainties on petrologic 310

data used for the calibration. Relative uncertainties on T are however much smaller, 311

in the range 10-15 °C (Beyssac et al., 2004).

312

Raman spectra were obtained using a Renishaw InVIA Reflex 313

microspectrometer (ENS Paris). We used a 514 nm Spectra Physics argon laser in 314

circular polarization. The laser was focused on the sample by a DMLM Leica 315

microscope with a 100 × objective (NA=0.85), and the laser power at the sample 316

surface was set around 1 mW. The Rayleigh diffusion was eliminated by edge filters, 317

and to achieve nearly confocal configuration the entrance slit was closed down to 10- 318

15 µm. The signal was finally dispersed using a 1800 gr/mm grating and analyzed by 319

a Peltier cooled RENCAM CCD detector. Before each session, the spectrometer was 320

calibrated with a silicon standard. Because Raman spectroscopy of CM can be 321

affected by several analytical mismatches, we followed closely the analytical and 322

fitting procedures described by Beyssac et al. (2002, 2003). Measurements were done 323

on polished thin sections cut perpendicularly to the main metamorphic rock fabrics 324

(mostly S1 or S2) and CM was systematically analyzed below a transparent adjacent 325

mineral, generally quartz. More than 15 spectra were recorded for each sample in the 326

static or extended scanning mode (1000-2000 cm

) with acquisition times varying 327

from 30 (static) to 60 - 150 (extended) seconds. Spectra were then processed using the 328

software Peakfit using a fitting procedure with 3 bands with Voigt profiles (Beyssac 329

et al., 2003; Beyssac and Lazzeri, 2012).

330 331

3.3. Petrology 332

The principal aim of this paper was to investigate metamorphic temperatures using 333

RSCM thermometry on a large scale. However we also felt it is important to place 334

these data in the context of previously available petrology, albeit without providing a 335

complete and parallel petrological assessment. Because published petrological data 336

was at least one, and in most cases three, decades old we undertook some new 337

analytical work and recalculations to update the available data.

338

Major element analyses of bulk-rock samples were carried out by X-ray 339

fluorescence spectrometry, with Fe

analyses by titration. Early results had been 340

obtained at Victoria University (Grapes et al., 1982). More recent results were 341

obtained at SpectraChem Analytical Ltd., Wellington, New Zealand, using methods 342

described in Vry et al. (2008), with Fe

analyses performed at the Albert-Ludwigs- 343

Universität, Freiburg, Germany, following the techniques outlined in Heinrichs and 344

Herrmann (1990). For these latter samples, the standard CRM2115 was also analysed,

345

and the result was Fe

= 8.54% compared to the stated value of Fe

= 8.50%; the 346

FeO determinations have a 1σ error of 0.08%. The powdered rock samples were 347

prepared at Victoria University using a TEMA tungsten carbide mill. All analytical 348

data are available from the PETLAB database (www.petlab.gns.cri.nz).

349

Electron probe microanalyses of rim compositions of garnet, plagioclase, and 350

coexisting matrix biotite were obtained by R. Grapes, mainly in the early 1990's 351

(Grapes and Watanabe, 1992; Grapes, 1995) using a JEOL JXA-733 SuperProbe, with 352

Moran Scientific software that incorporates modified ZAF and Bence-Albee matrix 353

corrections. The operating conditions used were 15 kV accelerating voltage, 12 nA 354

sample current, and a beam diameter of 1-3 µm. Some representative analyses are 355

given in supplementary material. All analysis positions were located using 356

backscattered electron imaging. Some representative data are given in supplementary 357

material and are available from the PETLAB database (www.petlab.gns.cri.nz).

358

To provide insights into the P-T conditions for the stability fields of relevant 359

mineral assemblages, P-T pseudosections were calculated in the 11 components 360

system MnNCKFMASHTO (MnO, Na

O, CaO, K

O, FeO, MgO, Al

O

, SiO

, H

O, 361

TiO

, O) using THERMOCALC v. 3.35 (Powell & Holland, 1988) and the internally 362

consistent thermodynamic dataset 5.5 (Holland & Powell, 1998). Mineral mixing 363

models and nonideality parameters are based on Holland & Powell (1998) and Powell 364

& Holland (1999). The activity-composition relationships used for garnet 365

(CaMnFMAS), white mica and paragonite (NKFMASH), plagioclase (NCAS), biotite 366

(KFMASHTO), epidote (CaFe

ASH), and chlorite (MnFMASH) are as described in 367

Vry et al. (2008). The activity-composition model for magnetite (FTO) is from White 368

et al. (2000), and the ilmenite (MnFTO) is from White et al. (2005), but with non 369

ideality described by W(ordered ilmenite, pyrophanite) = 2 kJ, W(disordered ilmenite, 370

pyrophanite) = 2 kJ, and W(hematite, pyrophanite) = 25 kJ (R. Powell, personal 371

communication, 2012). Albite, rutile, sphene, quartz, and H

O were treated as pure 372

end-members.

373 374

4. Results 375

376

4.1. Graphitic carbon in the Southern Alps

377

Graphitic carbon is widespread in rocks of the Southern Alps, generally dispersed in 378

the mineral matrix and totally absent only at a few localities. The latter localities may 379

correspond to unfavourable lithologies or lithologies affected by intense fluid 380

circulation which may be responsible for bleaching of carbonaceous material in the 381

rocks. In addition, distinctive hydrothermal graphite has been found in association 382

with orogenic gold-quartz mineral deposits in Otago (Pitcairn et al., 2005; Henne and 383

Craw, 2012; Hu et al. 2015). We found no such carbonaceous material in our samples 384

or fieldwork in the central Southern Alps. Figure 3 depicts representative Raman 385

spectra from the Southern Alps and demonstrates the general gradient of 386

graphitization following increasing metamorphism from SE to NW towards the 387

Alpine Fault. In the greywacke east of the drainage divide (Main Divide), 388

carbonaceous material exhibits Raman spectrum with several defect bands (e.g. D1, 389

D2, D3 and D4). Such spectra are characteristic of very disordered graphitic carbon 390

that were transformed under low-grade metamorphism at temperature below 330°C 391

(Beyssac et al., 2002; Lahfid et al., 2010). In this study, the peak temperature has been 392

assumed to be less than 330°C in these samples (Figure 4). To the west, there is a 393

progressive increase of the degree of graphitization through semischist and schist 394

towards the Alpine Fault. Close to the Main Divide, semischist samples exhibit 395

spectra with an intense main defect band as well as a strong D2 defect band as a 396

shoulder on the G peak: such spectra correspond to disordered graphitic carbon in 397

which the tridimensionnal aromatic skeleton remains poorly developed. Going 398

towards the Alpine Fault, both D1 and D2 bands decrease progressively with 399

increasing metamorphic grade and finally completely disappear in some of the highest 400

grade schist samples near the Alpine Fault. This shows a progressive graphitization 401

process that is completely achieved on the Alpine Fault. Importantly, detrital graphitic 402

carbon has been found in samples at all metamorphic grades. It can be easily 403

distinguished from in situ graphitizing organic matter based on: (i) morphological 404

criteria - as it generally appears as isolated grains or flakes; and (ii) Raman spectra - 405

as it usually exhibits a high crystallinity except in very high grade samples where it is 406

difficult to distinguish from organic matter from the simple observation of the Raman 407

spectra. Detrital graphite spectra were not included in RSCM T determinations. Note 408

that the presence of detrital graphite throughout the sequence has been observed in 409

many other metamorphic belts because graphite is easily recycled during the

410

erosion/weathering cycle (see Galy et al., 2008), and occurs widely in sedimentary 411

rocks.

412

More specifically, we have also carried out detailed investigation of inclusions 413

of graphitic carbon in the garnet porphyroblasts studied by Vry et al. (2004), sample 414

MA2 (VU37559 in Table 1). Graphitic carbon provides a marker of the garnet zoning 415

as it is present in some zones and absent from others, matching the chemical zonation 416

of garnets. The Raman spectrum of all such inclusions is constant and representative 417

of highly crystalline graphite. Calculating the temperature for such spectra yields ca.

418

575°C (Table 1) in good agreement with the maximum T obtained for the external rim 419

of garnet (ca. 600°C) and late Cenozoic ages. We conclude that all graphitic carbon in 420

this sample recorded the maximum T while garnet composition was only equilibrated 421

in the rims and not in the core during increasing metamorphism.

422 423

4.2. RSCM temperatures in the Southern Alps 424

All RSCM T are listed in Table 1 with key parameters such as location, geological 425

information, the number of spectra per sample, the mean R2 ratio parameter and T 426

with associated uncertainties. For very disordered graphitic carbon that is found in 427

least metamorphosed rocks, we assign T<330°C which is the lower bound of the 428

calibration by Beyssac et al., (2002). RSCM T are plotted on a regional-scale map 429

(Figure 4) and on three local maps depicting the main textural zones, metamorphic 430

mineral zone ‘isograd’ boundaries, faults and folds along the northern (Figure 5, 431

Wanganui to Taramakau), central (Figure 6, Copland to Whataroa) and southern 432

(Figure 7, Haast to Karangarua) segments of the Alpine Fault hanging wall. The latter 433

figures also include T profiles against the structural distance (D) to the Alpine Fault, 434

together with some approximate thermal field gradients drawn manually through the 435

data points. Curve fitting was deemed unwarranted for these transect gradients, due to 436

the relatively small numbers of samples involved in each transect. In addition, some 437

of the RSCM T data are depicted along four geological cross sections (Figure 8) 438

representing variations in geology along the Southern Alps: Waitaha river – Rakaia 439

valley, Whataroa river – Havelock river, Franz Josef – Godley valley, Karangarua - 440

Mt Cook village. Last, all RSCM T were plotted in frequency histograms for the 441

various metamorphic and textural zones (Figure 9).

442

443

Based on the dataset, these figures and Table 1, we make the following general 444

observations:

445

- Highest RSCM T measurements were from K-feldspar zone mylonites and schist 446

beside the Alpine Fault, where rocks locally contain only perfect graphite and are 447

inferred to have reached a minimum T of 640°C. Such very high T values are 448

observed in the Haast and Moeraki transects in the south, but also in Copland, 449

Waikukupa and Whataroa transects. There are two samples from garnet-oligoclase 450

zone rocks that contain only highly crystalline graphitic material yielding high T, 451

one from just south of Karangarua and the other at Wanganui river quarry.

452

- Lowest RSCM T measurements, where values have been assigned 330°C to 453

represent the current lower bound of the Beyssac et al. (2002) calibration were 454

observed in the vicinity of the Main Divide or to the southeast. Here the rocks are 455

TZ2a cleaved greywacke or textural zone TZ2b semischist, metamorphosed to 456

either sub-greenschist or chlorite zone.

457

- Plotting all RSCM T data versus metamorphic or textural zones shows the general 458

systematic T increase with increasing metamorphic grade and deformation (Figure 459

9). We note that some metamorphic (e.g. biotite or garnet-oligoclase) or textural 460

(e.g. TZ3 or TZ4) zones are characterized by a relative clustering of the T data, 461

whereas other zones (e.g. chlorite zone or textural zone 2B) exhibit a significantly 462

wider range of RSCM T measurements. This in part reflects the presence of some 463

relatively undeformed TZ2b rocks of the Aspiring lithologic association between 464

Waitaha-Arahura rivers, that are unusual in that they have reached amphibolite 465

facies metamorphism and yet retain remnants of original sedimentary structures 466

(Cox and Barrell, 2007; Cooper and Ireland, 2013).

467

- The RSCM T profiles exhibit a degree of similarity from the southernmost to the 468

northernmost transects. Plots of T as a function of D (Figures 5,6,7) nearly all 469

show RSCM thermal field gradients through the sub-greenschist to greenschist 470

facies (chlorite and biotite zone) rocks that are higher (>35°C/km) than field 471

gradients through amphibolite facies (garnet and K-feldspar zone) (<20°C/km).

472 473

In detail, the RSCM T data along the four geological sections of Figure 8 yields some 474

insight on the thermal evolution of the Alpine Schist. To the south, along the 475

Karangarua – Mt Cook village section (Figure 8d) there is extensive exposure of TZ3 476

and TZ4 rocks, corresponding to biotite and garnet zones, which comprise a map

477

thickness of more than 10 km. The rocks exhibit a fan-like structure, marked on a 478

broad scale by the opposite vergence of the Alpine Fault, which dips towards the SE, 479

and faults in the southeast, including the Main Divide Fault Zone (Cox and Findlay, 480

1985), which dip NW. This fan shape structure is mimicked by the main S3 481

crenulation cleavage and schistosity, which dips towards the SE close to the Alpine 482

Fault, is nearly vertical in the garnet zone and then progressively changes to dip 483

towards the NW in the garnet and biotite zones. The enveloping surface of folded, 484

early (S1/2) fabric and lithological variation dips gently southwest. Along this 485

geological section and nearby profiles (e.g. Copland or Karangarua), the field 486

gradients appear systematic with progressive increase in RSCM T towards the Alpine 487

Fault (Figures 6, 7). Yet in detail metamorphic sequences in pumpellyite-actinolite to 488

biotite zone rocks (TZ2a-3) near the Main Divide are locally inverted by juxtaposition 489

on NW-dipping faults (Cox et al., 1997; Craw, 1998) – a brittle juxtaposition which 490

results in steep thermal field gradients. Along the Moeraki profile there are also 491

kilometer-scale late- or post-metamorphic antiform and synform structures plunging 492

10-20° SW that fold the S2 surface (Rattenbury et al., 2010). Here strong T reversals 493

are present in RSCM measurements of garnet zone rocks (Figure 7). Other smaller 494

temperature reversals occur locally in data from the Otoko and Haast profiles, which 495

can also be attributed to late- or post-metamorphic folds (Figure 7; Cooper 1974;

496

Rattenbury et al., 2010). Where such folds are prominent, resulting thermal field 497

gradients are low. The Otoko profile also crosses an area of garnet-biotite-albite zone 498

samples within the greenschist facies, that contain very fine-grained (<1 mm) 499

grossular-spessartine garnets that are thought to be a remanent of Mesozoic 500

metamorphism exposed more widely in Otago (Figure 1; Mortimer, 2000; Rattenbury 501

et al., 2010). There do not appear to be any distinct steps or obvious thermal effects in 502

the dataset associated with this zone, which is distinguished by the ‘garnet-1 isograd’

503

in Figure 7.

504

In the Franz Josef – Fox area, the high-grade TZ3 and TZ4 schist units 505

metamorphosed to biotite zone and above, are much narrower (8 km, Figure 6), and 506

have been juxtaposed against semischist sequences by the BDTZ with escalator-like 507

component of dip-slip motion (Figures 6, 8c; Little et al., 2002b; Wightman and 508

Little, 2007). Here the distance between the Alpine Fault and the Main Divide is also 509

the smallest, being the region where the late Quaternary dip-slip rates are the greatest 510

on the Alpine Fault (>12 mm/year – Norris and Cooper, 2001), and uplift and

511

exhumation rates of schist also appear to be the highest along the plate boundary 512

(Little et al., 2005). The high grade schist units contain an S3 crenulation cleavage 513

dipping towards the SE, weakly oblique to the Alpine Fault, which steepens with 514

distance from the fault (Little et al., 2002a). The geological section across the schist 515

units contains upright SE plunging folds and a tight fan-like structure, with an abrupt 516

truncation of the structural trend towards the NW at the BDTZ. Along this section, 517

high RSCM T (graphitic material at >640°C) was observed in the Alpine Fault 518

mylonite zone, but RSCM T are relatively constant in the range 525-570°C in garnet 519

zone and biotite zone rocks structurally below the BDTZ (see Fox, Waikukupa, Franz 520

profiles on Figure 6). Structurally above the BDTZ, where foliation and bedding have 521

a predominantly NW dip, the thermal field gradients are much higher (>35°C/km, 522

potentially reaching ~90°C/km). Field gradients are also steep through chlorite and 523

biotite zone rocks of the Whataroa profile (Figure 6), which we interpret to be a 524

combination of the effects of topography, flat lying S2 fabric, and fault juxtaposition 525

of different metamorphic blocks (Figure 8b).

526

Features observed in the southern and central profiles are also present in the 527

northern region (Figures 5, 8a,b). The high-grade TZ3 and TZ4 schist units widen to 528

14 km in both map view and geological section, and there are a number of mapped 529

kilometer-scale late- or post-metamorphic antiform and synform structures which 530

appear to fold isotect and isograd boundaries between Wanganui and Hokitika rivers.

531

Such folds are relatively tight with steep limbs in foliated schist sequences, but 532

typically open in the semischists (Figure 8a). The result appears to be a relatively flat- 533

lying S2 enveloping surface, similar to the Whataroa-Havelock section (Figure 8b), 534

plunging gently-moderately to the northeast, which produces temperature reversals in 535

the Waitaha profile (Figure 8a). TZ2a and TZ2b semischist sequences (greenschist 536

facies) are relatively flat-lying, with structural blocks juxtaposed by subhorizontal 537

folding or faults (Andrews et al., 1974). Thermal field gradients are high (>45°C/km) 538

in the sub-greenschist to biotite zone rocks, and very low (<10°C/km) across schist 539

sequences.

540 541

4.3. Petrological constraints 542

Petrological data were collated for samples from the Franz Josef-Fox area, including 543

older studies (Grapes and Watanabe, 1992; Grapes, 1995). Traditional garnet-biotite 544

geothermometry, and temperature estimates based on results of the P-T pseudosection

545

calculations new for this study are presented in Table 2. An example of pseudosection 546

calculation is given as supplementary material. These data are shown on Figure 10 547

along with P-T estimates extracted from Figure 4 in Grapes and Watanabe (1992), and 548

shown in map view on Figure 11. Note that garnet-biotite temperatures in our study 549

are generally higher than those from Grapes and Watanabe (1992), based on the same 550

mineral analyses. In our study, we used the calibration by Hodges and Spear (1982) 551

based on rim analyses using pressure estimates based on results of garnet-biotite- 552

muscovite-plagioclase barometry (Hoisch, 1990, Fe-endmember). Grapes and 553

Watanabe (1992) used the calibration by Ferry and Spear (1978), as modified for Ca 554

content by Hodges and Spear (1982) and Hoinkes (1986), and the garnet-biotite- 555

muscovite-plagioclase geobarometer of Ghent and Stout (1981), with modification by 556

Hodges and Crowley (1985). The P and T uncertainties (± 1 kbar and ± 50°C; Grapes 557

and Watanabe, 1992), as estimated from standard deviations, apply for both studies.

558 559

5. Discussion 560

561

5.1. Comparison of RSCM data with petrological constraints 562

At some localities close to the Alpine Fault, in the high-temperature range, there is 563

good agreement between RSCM thermometry and the peak T estimated by petrology.

564

Nearly pure crystalline graphite is present yielding RSCM T above 580°C close to the 565

Alpine Fault, and in many places perfect graphite is found in the central and southern 566

areas yielding RSCM T above 640°C, the higher limit of the calibration by Beyssac et 567

al. (2002). This is in agreement with the T estimates by Vry et al. (2004) who showed 568

that the maximum T recorded by a garnet porphyroblast in Hokitika area (Mac’s 569

Creek) is ca. 600°C using garnet-biotite thermometry. The lowest RSCM T occur 570

from the Main Divide southeastwards, where most samples have been assigned 571

<330°C based on the lower limit of the Beyssac et al. (2002) calibration. There are no 572

quantitative petrological estimates of T for quartzofeldspathic lithologies 573

metamorphosed to such low temperatures. However the observed low temperatures 574

are in agreement with: 1) the observed metamorphism of rocks at prehnite- 575

pumpellyite and pumpellyite-actinolite facies. For instance, prehnite occurring 576

together with pumpellyite typically indicates temperatures in the range 250-300 °C 577

and pressures below about 2 kbar (e.g. Willner et al., 2013); 2) the presence of

578

partially annealed fission tracks in zircons, with zircon ages which suggest rocks had 579

exceeded 240°C closure temperatures and have been uplifted from partial annealing 580

zone or deeper (Batt et al., 2000; Cox and Findlay 1995; Tippett and Kamp 1993a,b);

581

3) K-Ar ages that suggest the rocks have partial retention of gas and remained below 582

temperatures of ca. 300°C (Batt et al., 2000).

583

All methods record the decrease in peak temperature away from the Alpine 584

Fault towards the southeast (Figure 11), but RSCM T are generally higher than any 585

related estimate from petrology and yield significantly lower apparent field 586

metamorphic gradients through the high-grade schist. This is most-clearly illustrated 587

on Figure 10 where all RSCM T data for the southern, central and northern profiles 588

are shown together with petrological T estimates from Grapes and Watanabe (1992) 589

and this study (see Table 2) from the Franz Josef – Fox area. The T estimates by 590

Grapes and Watanabe (1992) are definitely lower by several tens of degrees compared 591

to RSCM T (central Southern Alps profiles) except those in the direct vicinity of the 592

Alpine Fault where they converge towards ca. 600°C. THERMOCALC T estimates 593

are higher than those of Grapes and Watanabe (1992) and for the most-part are lower 594

than RSCM T.

595

A possible explanation of the difference between RSCM T and petrological 596

estimates is due the irreversibility of graphitization (Beyssac et al. 2002) such that 597

RSCM records the peak T and is not sensitive to the retrograde/exhumation path of 598

the rocks. By way of contrast, mineral assemblages can re-equilibrate their chemistry 599

during retrogression, modifying and erasing the peak T signal, especially when fluid 600

circulation and deformation are important. In the Alpine Schist, garnet compositions 601

are not simple Fe-Mg end-member mixtures, and can vary considerably from rock to 602

rock, and the biotite can be subject to re-equilibration and regrowth during uplift and 603

cooling. Fluid flow has locally affected some of the rocks, and the effects and timing 604

of this may pass unrecognized, and have not been quantified. The MnO contents in 605

the bulk rock compositions vary, and the first appearance of garnet in P-T 606

pseudosections is very sensitive to this, as well as the choices of activity-composition 607

models. In any case, we note that temperature estimates based on results of P-T 608

pseudosection calculations for rock samples that contain relatively high-grade mineral 609

assemblages (containing ilmenite, oligoclase, ± garnet), are typically only slightly 610

lower than, and within error of, the RSCM T obtained from nearby samples.

611

However, for the samples that contain lower-grade mineral assemblages with sphene,

612

there is a larger temperature difference, with higher T estimates being obtained from 613

the RSCM data (see Figure 10).

614

At some localities (see Table 2), the discrepancy between RSCM T and 615

petrological T estimates is, nonetheless, somewhat surprising. RSCM thermometry 616

has now been used in many various geological contexts and generally exhibits good to 617

excellent concordance with conventional petrology, including garnet-biotite 618

thermometry (e.g. see Plunder et al., 2012 and references therein). However, in most 619

available studies, RSCM T was applied to simple thermal histories, i.e.

620

monometamorphic history with one single thermal event, except the notable examples 621

of some internal units of Taiwan (Beyssac et al. 2007). But given the high quality and 622

consistency of the RSCM spectral data in the Alpine Schist, and considering that 623

minerals may not record the true peak metamorphic temperature, we consider that 624

RSCM thermometry yields a first quantification on a large scale of peak metamorphic 625

626 T.

627

5.2. RSCM T pattern in the Southern Alps 628

Figure 10 is a compilation of all local-scale profiles presented on Figures 5, 6 and 7 629

for southern, central and northern Alpine Schist. RSCM T data from all metamorphic 630

zones has also been projected onto an along-strike section parallel to the Alpine Fault 631

(Figure 12). Also shown are the approximate position of key metamorphic mineral 632

‘isograds’ (first appearance of biotite; garnet and K-feldspar), and the zone where 633

young (<6 Ma)

Ar-

Ar and K/Ar ages have been obtained in the central Southern 634

Alps. It is useful compare the along-strike projection against Figure 10 that shows the 635

same data plotted perpendicular to the Alpine Fault. From Figures 10 and 12, it seems 636

there may be a decrease in peak T along the Alpine Fault in the K-feldspar zone going 637

towards the northeast, although this trend may in part be apparent due to the absence 638

of samples analysed from the northernmost area. Both the garnet-oligoclase and 639

biotite zones exhibit a relatively clustered pattern of peak T around 560°C and 520°C 640

respectively all along strike. There is dispersion in both cases around the mean values 641

for each zone especially in the central Southern Alps although this may also be due to 642

a denser sampling bias in this area. The chlorite zone all along the Alpine Fault covers 643

an extremely wide range of T from ca. 360°C reaching as high as ca. 550°C. This is a 644

very wide T range for a classical chlorite greenschist facies zone, especially towards 645

the high T. One possibility is that the lithology/bulk chemistry may not have allowed

646

growth of index minerals (garnet and/or biotite, see Vry et al., 2008), or the rocks 647

suffered complete retrogression obliterating some high-grade minerals, resulting in 648

the possibility that samples could have been ‘misclassified’ as being apparently of 649

lower metamorphic grade than the temperature they actually experienced.

650

Figure 10 shows that higher RSCM T extends farther from the Alpine Fault 651

along the southern and, to a lesser degree, the northern compared to the central 652

profiles. This is due to the wider map extension of high-grade units (biotite and garnet 653

oligoclase) in the south and north, whereas these units are far less thick and steeper in 654

the central Southern Alps. Although the data are somewhat scattered, it seems that the 655

RSCM T pattern along both the central and northern profiles has nearly constant high 656

T within the first kilometers close to the Alpine Fault (ca. 6 km for the central and ca.

657

8 km for the north) followed by a steep decrease of T in the eastern biotite and 658

chlorite zones. In the south, the T profile is more continuous and progressive, with no 659

break in the slope, and projection toward Alpine Fault suggest rocks here may have 660

reached temperatures ~700°C. This is consistent with widespread occurrences of 661

pegmatite in the Mataketake Range and less commonly elsewhere between Haast and 662

Moeraki rivers, which reflect partial melting of the schist that occurred during the 663

Late Cretaceous (Chamberlain et al., 1995). The calculation of D for these data 664

assumes the Alpine Fault has a constant dip of 45° (Figure 2). Also shown in Figure 665

10 are the range of possible D values and field gradients that could occur if the fault 666

dipped as shallow as 30° or as steeply as 60°.

667 668

5.3. Age of RSCM temperatures in the Southern Alps 669

RSCM thermometry records the peak metamorphic T undergone by carbonaceous 670

material and the host rock during the burial history but it carries no age information 671

by itself. In the case of complex poly-phased metamorphic histories like in the Alpine 672

Schist with at least two major thermal events, either associated with tectonics on the 673

margin of Gondwana during Jurassic – Cretaceous or with evolution of the present 674

plate boundary during late Cenozoic, an important issue remains with regards to the 675

age of the recorded RSCM T. This is a significant question if one wants to use such 676

data to infer and constrain models of the thermal structure of the Southern Alps, or 677

rheology of Alpine Fault rocks at depth (e.g. Toy et al. 2010), as the T measurements 678

will be most directly relevant if they are late Cenozoic in age. Confirming the RSCM 679

T measurements as late Cenozoic is made more difficult by the complete absence of

680

Cenozoic-aged metasedimentary units in the Southern Alps that would only contain 681

the expression of late Cenozoic metamorphism. Age of mineral assemblages and of 682

peak P-T conditions recorded in the Alpine Schist have therefore long been discussed 683

in the literature (e.g. Chamberlain et al., 1995; Little et al. 2002a,b, 2005; Vry et al.

684

2004, 2008), but there is no real consensus as to the exact locations where peak P-T 685

conditions recorded by the rocks are late Cenozoic. There are a number of arguments 686

however that supports this to be the case, at least locally, for Alpine Schist between 687

Fox and Franz Josef, and potentially Copland – Whataroa rivers, as we discuss below.

688

On a large scale, the spatial distribution of RSCM T is strongly correlated with 689

the geometry of the main contacts and tectono-stratigraphic structures that were 690

acquired during formation of the Southern Alps. But the classical two-dimensional 691

model of exhumation by Wellman (1979), exhuming deep old rocks up a ramp formed 692

by the Alpine Fault, would yield the same results whether the peak T was late 693

Cenozoic, or a product of older Jurassic-Cretaceous metamorphism with passive 694

exhumation up the Alpine Fault ramp during the late Cenozoic. Instead, 695

thermochronologic data at least record when the rocks have cooled below some 696

threshold temperature (closure temperature), depending on the selected 697

thermochronometer. Except for U-Pb geochronology on zircon yielding 698

crystallization ages, the highest temperature thermochronometer so far applied has 699

been

Ar-

Ar ages on hornblende. Between Franz Josef and Fox, hornblende from 700

mylonite beside the Alpine Fault has

Ar-

Ar ages older than Neogene formation of 701

the plate boundary, yet garnet-oligoclase zone Alpine Schist been reset to < 6 Ma 702

when the Southern Alps were uplifted (Figures 4, 11; Chamberlain et al., 1995; Little 703

et al., 2005). The reset ages are distributed through a zone where measured RSCM T 704

range between 530-565°C (Figure 11). These RSCM temperatures are therefore 705

equivalent to, or higher, than the commonly assumed value for the closure 706

temperature of

Ar-

Ar ages on hornblende which is ca. 500-550°C. At least, in this 707

zone it seems reasonable to assume that RSCM T represents a peak T which was 708

reached during the late Cenozoic. Further support is given by the close 709

correspondence between (i) peak T of ca. 600°C recorded by the most external rim of 710

garnet porphyroblasts dated to Cenozoic ages (Vry et al., 2004) and (ii) RSCM T of 711

ca. 575°C obtained on all graphitic inclusions on the same garnet from the core to the 712

rim.

713