Inverted distribution of ductile deformation in the relatively “dry” middle crust across the Woodroffe Thrust, central Australia

Thrust fault systems typically distribute shear strain preferentially into the hanging wall rather than the footwall. The Woodroffe Thrust in the Musgrave Block of central Australia is a regional-scale example that does not fit this model. It developed due to intracontinental shortening during the Petermann Orogeny (ca. 560-520 Ma) and is interpreted to be at least 10 600 km long in its E-W strike direction, with an approximate top-to-north minimum displacement of 60-100 km. The associated mylonite zone is most broadly developed in the footwall. The immediate hanging wall was only marginally involved in the mylonitization process, as can be demonstrated from the contrasting thorium signatures of mylonites derived from the upper amphibolite facies footwall and the granulite facies hanging wall protoliths. Thermal weakening cannot account for such an inverse deformation gradient, as syn-deformational P-T estimates for the Petermann Orogeny in the hanging wall and footwall 15 from the same locality are very similar. The distribution of pseudotachylytes, which acted as preferred nucleation sites for shear deformation, also cannot provide an explanation, since these fault rocks are especially prevalent in the immediate hanging wall. The most likely reason for the inverted deformation gradient across the Woodroffe Thrust is water-assisted weakening due to the increased, but still limited, presence of aqueous fluids in the footwall. We also infer that the relative increase of fluids in the footwall in the direction of thrusting, together with the slight decrease in the temperature of mylonitization (ca. 20 100 °C), was responsible for the observed 6-fold decrease in thickness (from ca. 600 m to 100 m) of the Woodroffe Thrust mylonitic zone.


Introduction
Continental fault systems with displacements on the order of several tens to hundreds of kilometres generally show an asymmetric mylonite distribution across the main fault horizon that is opposite for reverse faults or thrusts and normal faults or detachments. Fault zones are predicted to become more viscous and broaden with depth (e.g., Fossen and Cavalcante, 2017; than the upper boundary (Fig. 4b). The thickness of the mylonitic zone at location 1 ( Fig. 1) was not calculated, because the Woodroffe Thrust is only exposed along strike. The results for all other studied transects are summarized in Table 1   observation that hanging wall rocks are depleted in thorium compared to footwall rocks (Fig. 5). This contrast is due to (1) dehydration and melting reactions during the earlier ca. 1200 Ma Musgravian Orogeny, which depleted the granulite facies hanging wall to a greater degree than the upper amphibolite facies footwall (Heier and Adams, 1965;Heier, 1967, 155 1968), (2) the predominance of granitoids rather than gneisses in the footwall, and (3) the fact that low-thorium Giles Complex gabbro-noritic intrusions are often exposed in the immediate hanging wall of the Woodroffe Thrust. Deformation and metamorphism during mylonitization did not significantly alter the original thorium content of the rocks, since the thoriumbearing phases, such as zircon, allanite, monazite and apatite did not break down during the Petermann Orogeny. Consequently, the original variation in thorium between the granulite and amphibolite facies rocks was preserved (Fig. 5). Anomalies in this 160 broad pattern are present locally and can be attributed to hanging wall-derived alluvial sediments in the immediate footwall (lower-thorium anomaly) and to granitoid intrusions of the Pitjantjatjara Supersuite and Giles Complex (higher-thorium anomaly), which are syn-to post-Musgravian upper amphibolite to granulite facies metamorphism Young et al., 2002) (Fig. 5). The contrast between a lower-thorium hanging wall and a higher-thorium footwall is welldefined on the airborne thorium map for the central and southern locations 6, 8 and 11-14, but less evident in the northern 165 locations 1-5 and 7 (Fig. 5). Thorium measurements were carried out on thin section chips of felsic gneisses and granitoids via γ-ray spectrometry. The method is outlined in detail in the Supporting Information S1.

Thorium concentration in felsic units
Thorium contents of felsic gneisses and granitoids (Table 2) have been compiled from the current study as well as from previous studies of nearby areas (Camacho, 1997;Young et al., 2002). These data have been grouped into (1) the central and 170 southern locations (6, 8, and 11-14) and (2) the northern locations (1-5, and 7), based on the airborne thorium map, as introduced above.
In group (1), samples unequivocally attributed to either the hanging wall or footwall of the Woodroffe Thrust were grouped together. This was done on the basis of their geographical position, with respective samples originating either from far into the non-mylonitized hanging wall and footwall or from the lowermost protomylonitic part of the Woodroffe Thrust. In the hanging 175 wall, thorium concentrations range from 1 to 28 ppm in felsic gneisses and up to 63 ppm in granitic intrusions (higher-thorium anomalies), but are in both cases usually lower than 8 ppm. In the footwall, concentrations vary between 2 and 195 ppm and are typically higher than 10 ppm (Table 2). These concentrations are in accord with the regional-scale contrast in thorium concentrations across the Woodroffe Thrust in the central and southern locations 6, 8 and 11-14 (Fig. 5), as well as with the results of Lambert and Heier (1968), who determined concentrations of 2.1 ppm for the granulite facies hanging wall and 11 180 ppm for the upper amphibolite facies footwall. Based on the compilation in Table 2

Determination of the hanging wall-footwall boundary
Measured thorium concentrations indicate that in locations 8 and 11-14 (Fig. 5) the uppermost mylonites of the Woodroffe Thrust developed in the lower-thorium Fregon Subdomain (Table 2). These results are in agreement with the proposed 205 identification of hanging wall reworking based on field appearance. Similar field relationships, such as progressive downwards mylonitization of hanging wall pseudotachylyte breccia, also indicates limited reworking of the Fregon Subdomain at locations 4 and 6 ( Fig. 5). However, the lack of a clear contrast in thorium concentrations between footwall and hanging wall in the northern locations precludes any verification of these field observations, as well as any independent determination of the thickness of hanging wall-derived mylonites in the thrust zone. We therefore excluded these northern transects when using 210 thorium concentration as a proxy to map the original boundary between the hanging wall and footwall (Fig. 6). It was also not possible to precisely determine the boundary at location 13 ( Fig. 1), due to a large lateral gap between the last sample assigned to the footwall and the first sample assigned to the hanging wall. In an attempt to quantitatively calculate the degree of hanging wall/footwall reworking into the total mylonite zone, we applied trigonometry based on the geometrical arrangement sketched in Fig. 4a. The only modification was that the lower mylonite boundary was now defined by the newly reconstructed boundary 215 between hanging wall and footwall. Results are summarized in Table 3. In contrast to the 250 m proposed by Bell (1978) for the Amata area (western edge of Fig. 1), our results indicate that only the lowermost 3 m of the Fregon Subdomain were reworked into the Woodroffe Thrust mylonites at location 8, increasing up to 18-40 m at locations 11-14 (Fig. 1). These values represent 3-6 % of the entire thickness of the Woodroffe Thrust mylonites (Table 1) at locations 8, 11 and 14 and up to 13 % at location 12 ( Fig. 1). However, the difference in elevation (p) is not well defined over short distances (q) given the limited 220 (30x30 m) resolution of the digital elevation model (ASTER). This can introduce a significant uncertainty into the calculation of the degree of hanging wall reworking. Nevertheless, our analysis clearly shows that the majority of mylonites developed in the Mulga Park Subdomain (footwall) rather than the Fregon Subdomain (hanging wall).

Introduction of carbon
Finely dispersed calcite is locally found with very low modal abundance (typically <1 %) in the otherwise non-carbonaceous rocks of the central Musgrave Block (Fig. 7c). Calcite-bearing samples are present throughout the study area, but are generally more common towards the north (Fig. 8b). Microstructures indicate that the fine-grained (<10 µm) calcite nucleated during shearing (Fig. 7c). Calcium was made available from recrystallization of plagioclase to new grains with lower anorthite content 245 , but carbon requires an external fluid since the protoliths are entirely non-carbonaceous. In the attempt to establish the origin of this fluid, carbon and oxygen isotopes of calcite were measured (Supporting Information S2), yielding mean average values of -4.1 ‰ for δ 13 CCal (V-PDB) and +10.1 ‰ for δ 18 OCal (SMOW). Within the same samples, the whole rock isotopic signature δ 13 Cwhole rock (V-PDB) is always lower (on average by 2.7 ‰) than the corresponding δ 13 CCal values.

Plagioclase stability and breakdown
Plagioclase recrystallized in the Woodroffe Thrust mylonites and associated shear zones (Bell and Johnson, 1989;Wex et al., 2017) forming typical "core-and-mantle structures". Mineral inclusions within plagioclase clasts are common and allow the distinction of four different types of clasts, respectively termed microstructures 1 to 4: (1) Plagioclase studded with abundant inclusions of epidote and muscovite (Fig. 7d). This type is, with one exception, 270 restricted to the northern locations 1-9 (Fig. 8c).

Abundance of hydrous minerals
The modal abundance of hydrous minerals in deformed and undeformed pseudotachylytes (representative compilation in Supporting Information S3) from felsic footwall and hanging wall samples was determined by image analysis of backscattered electron (BSE) images (method outlined in the Supporting Information S1). The different amounts of hydrous minerals 290 between the pseudotachylytes should reflect compositional variations between the host rocks from which they formed. The results, summarized in Table 4, indicate that matrix mineral assemblages of felsic pseudotachylytes from the hanging wall and footwall are similar to each other. These assemblages dominantly comprise Pl + Kfs + Qz + Bt + Mag + Ilm with individual samples also containing Ep, Grt, Cpx, Opx, Ky, Ms, Rt or Hbl. However, there is a strong contrast with regard to the modal abundance of hydrous minerals. Samples have been sorted into northern (locations 1-5 in Fig. 1), central (locations 6-9 in Fig.  295 1) and southern location groups (locations 11-14 in Fig. 1), revealing that: (i) the abundance of hydrous minerals decreases from north to south in both the hanging wall and the footwall; and (ii) at similar latitude, the footwall rocks are more hydrous than the hanging wall rocks. Two samples (SW14-029A and SW14-179) do not fit this regional trend and these outliers have
For the sake of simplicity, none of these reactions consider the presence of iron. However, iron is certainly necessary to account for the crystallization of epidote and was potentially derived from relict iron-bearing minerals (e.g., magnetite or ilmenite). 320 Microstructure 1 is unequivocally correlated with hydration reaction 1. Reaction 2 is capable of producing microstructure 2, but garnet has been interpreted to serve as a calcium-sink rather than source during the high-pressure Petermann Orogeny (Camacho et al., 2009). Alternatively, we propose that the decrease in anorthite content during plagioclase recrystallization  provided the necessary source of calcium for producing microstructure 2. Microstructure 3 is correlated with reaction 3. Reactions 3 and 4 represent the high-pressure breakdown of plagioclase under anhydrous and hydrous 325 conditions, respectively. According to Wayte et al. (1989), the transition between the two competing reactions occurs at a water activity of ca. 0.004 for pressure and temperature conditions similar to those in the more southerly footwall locations. In samples with microstructure 3, epidote is never observed as a secondary inclusion phase together with kyanite (Supporting Information S4), indicating that reaction 3 was always preferred over reaction 4.

Fluid activity 330
In the southern locations 13-15 (Fig. 1), plagioclase breakdown by reaction 3 indicates that water was not sufficiently available to drive reaction 4 (Goldsmith, 1980(Goldsmith, , 1982Wayte et al., 1989). Consequently, reaction 3 is a good indicator for very low water activities (<0.004; Wayte et al., 1989) and the absence of a free aqueous fluid phase during deformation, since only very small amounts of water (ca. 20 ppm) are required for mineral reactions in a solid silicate system (Milke et al., 2013). Similarly, relict plagioclase clasts without any inclusions (microstructure 4) also indicate the absence of free aqueous fluids, as the studied 335 rocks were all metastable with respect to hydration reactions 1, 2 and 4 ( Fig. 9). Consequently, any of these breakdown reactions would have rapidly consumed any available free fluid, since all other reactants were present in the studied samples.
Vice versa, plagioclase breakdown by reaction 1 clearly indicates the presence of a free aqueous phase during deformation.
Reaction 2 also involves hydration, as indicated by the marginal presence of epidote in plagioclase clasts (microstructure 2).
However, free aqueous fluids were not sufficiently available to facilitate reaction 1. Hence, we favor the interpretation that 340 microstructure 2 indicates largely anhydrous conditions, with only very minor fluid introduction.
Based on the classification above, the regional availability of a free aqueous fluid phase during deformation, respectively termed "wet" and "dry", can be determined from the mineral inclusions in plagioclase (Fig. 8c) and the corresponding inferred breakdown reactions (Sect. 8.1). The distinction is purely qualitative and refers only to whether or not free aqueous fluid was sufficiently available to facilitate the hydrous breakdown reaction 1. Consequently, the studied field area is characterized as 345 dominantly "dry" (microstructures 2, 3 and 4) with a progression towards locally "wet" conditions (microstructure 1) in most of the northernmost exposures (Fig. 10). The regional variation in (1) the development of syntectonic quartz veins (Fig. 8a) and (2) the introduction of carbon (Fig. 8b) are each consistent with this interpretation (Fig. 10). However, within a single location, individual samples can have a "wet" microstructure 1 while others still preserve a "dry" microstructure 4 (Fig. 8c), indicating that the fluids were only present on a very local scale. This is additionally supported by the coeval development of 350 Solid Earth Discuss., https://doi.org/10.5194/se-2018-9 Manuscript under review for journal Solid Earth Discussion started: 7 February 2018 c Author(s) 2018. CC BY 4.0 License. a hydrous (Pl + Opx + Grt + Hbl + Ilm + Mag ± Cpx ± Bt) and an anhydrous mineral assemblage (Pl + Opx + Grt + Cpx + Kfs + Qz + Ilm + Mag) within a single thin section of a sheared dolerite dyke from the footwall of the Woodroffe Thrust (Supporting Information S5).

Fluid source
Based on the above quantification of fluid activity in the study area, it is evident that aqueous fluids and CO2-dominated brines 360 were introduced in the northern exposures of the Woodroffe Thrust (Fig. 10). However, there does not seem to be any obvious link between the infiltration of the aqueous fluids and that of the CO2-dominated brines, since samples that preserve a "wet" microstructure 1 are not necessarily calcite-bearing and vice versa (Figs. 8b,c).

Aqueous fluids
Aqueous fluids infiltrating into the footwall of the Woodroffe Thrust were unlikely to have been derived from the "dry" 365 hanging wall (Table 4) and consequently, must originate from units within or underlying the current level of exposure of the footwall. Gneisses and granitoids in the "wet" northern part of the study area are clearly interleaved with and juxtaposed onto the basal units of the Amadeus Basin (Wells et al., 1970), as evident from outcrops of Dean Quartzite (Forman, 1965;Young et al., 2002) immediately west of the Kelly Hills klippe (locations 1-5 in Fig. 1). These Neoproterozoic sedimentary rocks represent an ideal source for aqueous fluids since they were metamorphosed and dehydrated during the Petermann Orogeny 370 (Wells et al., 1970). However, based on regional-scale geometric reconstructions, it has been argued that the Woodroffe Thrust and potentially underlying thrust planes developed in-sequence . If this is true, then the Dean Quartzite was only imbricated below the northernmost studied locations after movement and deformation on the Woodroffe Thrust had largely ceased and the dynamic microstructures had already been frozen in. Therefore, we prefer a model in which the aqueous fluids were released from the granitoids and upper amphibolite facies gneisses within the footwall. Such a model is in 375 agreement with the fact that the regional trend towards higher abundance of hydrous minerals in the north parallels the shift towards "wet" conditions, as indicated by the plagioclase breakdown reactions (Fig. 10). An internal source is also consistent with the conclusion that "wet" conditions were only present on a very local scale and that "wet" and "dry" samples are often preserved in close proximity. Consequently, we consider that the studied field area was not infiltrated by externally derived fluids and remained a relatively closed system. distribution of ductile deformation, as they are significantly more abundant in the hanging wall compared to the footwall (Camacho et al., 1995;Lin et al., 2005). 420 A potential explanation for the observed inverse gradient of ductile deformation along the Woodroffe Thrust is water-assisted weakening (e.g., Griggs, 1967Griggs, , 1974Griggs and Blacic, 1965;Hobbs, 1985;Kronenberg et al., 1990;Stünitz et al., 2017;Tullis and Yund, 1989). As discussed in Sect. 8.3.1, the rocks in the study area represent a relatively closed system with respect to the presence or absence of aqueous fluids. Hence, the availability of aqueous fluids is directly linked to the abundance of hydrous minerals in these rocks, which is observed to decrease from north to south (Fig. 10). Similarly, the contrasting 425 abundance of hydrous minerals in hanging wall and footwall (Table 4) indicates that there is consistently a higher potential for generating aqueous fluids in the footwall than in the hanging wall. Overall, the majority of the studied rocks are relatively "dry", so that only slight differences in the abundance of water could facilitate an asymmetric localization of ductile deformation as a consequence of water-assisted weakening. The preferential mylonitization of the Mulga Park relative to the Fregon Subdomain (Fig. 6) therefore seems to be largely controlled by the precursor mineralogy established as the result of 430 the earlier (ca. 1200 Ma) Musgravian Orogeny metamorphism, with peak conditions of upper amphibolite facies in the footwall but granulite facies in the hanging wall. This supports the initial hypothesis of Bell and Etheridge (1976) and Camacho et al.
(1995) that a more hydrous footwall underlying an anhydrous hanging wall can facilitate an inverted gradient of ductile deformation.
These arguments may explain the overall preferential localization of deformation in the Mulga Park Subdomain, but fail to 435 provide an adequate explanation for why the lowermost Fregon Subdomain was also incorporated into the mylonites (Figs. 2,   6, 10). The generally low abundance of hydrous minerals in both hanging wall and footwall in the "dry" southern locations (11-14 in Table 4) potentially promoted a similar rheological response in both units. Locally, there may be no contrast at all in protolith composition, which might explain the marginal mylonitization of the lowermost Fregon Subdomain. Reworking of the hanging wall may also have been guided by the presence of pseudotachylytes, which are ubiquitous and voluminous in the 440 immediate hanging wall of the Woodroffe Thrust (Camacho et al., 1995;Lin et al., 2005).

Variation in mylonite thickness
Mylonite thickness in the study area does not appear to significantly vary parallel to strike (i.e. E-W). However, from south to north, i.e. parallel to the direction of tectonic transport, the Woodroffe Thrust mylonitic zone gradually decreases in thickness from over 600 m to less than 100 m (Table 1), indicating a 6-fold increase in shear strain (and therefore average strain rate) 445 within the mylonites, assuming that the relative displacement across the Woodroffe Thrust was constant along the entire ca. 60 km transect. This trend from south to north is accompanied by a slight decrease in metamorphic temperature of ca. 100 °C  and the shift from "dry" to "wet" conditions, as reflected in the increasing abundance of hydrous minerals (Fig. 10). The thickness of large-scale shear zones is considered to decrease with decreasing temperature and depth (Platt and Solid Earth Discuss., https://doi.org/10.5194/se-2018-9 Manuscript under review for journal Solid Earth Discussion started: 7 February 2018 c Author(s) 2018. CC BY 4.0 License.
Behr, 2011a, 2011b) and displacement (Hull, 1988), but may also decrease with increasing fluid-rock interaction as a 450 consequence of volume loss (e.g., Newman and Mitra, 1993). A point in the footwall only enters the shear zone when it passes the toe of the thrust (that is where the thrust meets the surface). This could potentially lead to a variation in the finite shear strain experienced in the footwall. However, the whole of the exposed ca. 60 km N-S section was formerly at mid-to lower crustal level  and thus nowhere near the thrust toe. There is also currently no evidence for major splays of the Woodroffe Thrust into the hanging wall. We therefore assume that the studied section experienced more or less the same 455 relative displacement and that variation in this parameter cannot account for the difference in mylonite thickness. It was argued above that the greater abundance of hydrous minerals in the footwall Mulga Park Subdomain compared to the hanging wall Fregon Subdomain, reflecting peak metamorphic conditions of upper amphibolite facies and granulite facies respectively during the earlier Musgravian Orogeny, resulted in slightly "wetter" conditions in the footwall during activity of the Woodroffe Thrust. This was proposed as an explanation for the broader zone of mylonites in the footwall, which in turn would argue 460 against the increasingly hydrous conditions toward the north as an explanation for the decrease in overall mylonite width. The general decrease in mylonite zone thickness toward the north is therefore interpreted to be due to the established decrease in metamorphic temperature, promoting localization in a narrower zone.

Conclusions
Field and thin section observations establish that the rocks of the central Musgrave Block were predominantly "dry" during 465 development of the midcrustal Woodroffe Thrust during the ca. 560-520 Ma Petermann Orogeny, but with a progression in the thrust direction towards locally "wet" conditions in some of the northernmost exposures. This is indicated by: (1) rare occurrence of syntectonic quartz veins and quartz-rich pegmatites (locally found only in the north), (2) metastability of plagioclase in the presence of K-feldspar, which rarely shows significant sericitization via the reaction An + Or + H2O = Ms + Ep + Qz (more common towards the north) and (3) preferential high-pressure breakdown of plagioclase via the reaction An 470 = Grs + Ky + Qz (common in the southerly exposures), rather than An + H2O = Ep + Ky + Qz. Aqueous fluids were most likely derived internally from hydrous minerals within the footwall gneisses and granitoids, implying that the rocks in the study area were a relatively closed system.
The thickness of the Woodroffe Thrust mylonites generally increases with increasing metamorphic grade and does not appear to be linked to the presence or absence of an aqueous fluid. However, atypical of a thrust, ductile deformation is more 475 extensively developed in the footwall rocks and only marginally involved several tens of metres of the lowermost hanging wall. The inverse gradient of ductile deformation cannot be explained by thermal weakening or the distributed presence of pseudotachylyte (acting as preferred nucleation sites for shearing), but rather by preferential water-assisted weakening in the "wetter" footwall compared to the "dry" hanging wall. This reflects the earlier (Musgravian Orogeny) peak metamorphic conditions (granulite facies in the hanging wall and upper amphibolite facies in the footwall) and the contrasting availability 480 of aqueous fluids derived from relict hydrous minerals in the footwall and hanging wall.