Evolution of fluid chemistry and fluid-flow pathways during folding and faulting: an example from Taemas, NSW, Australia

Abstract In the Taemas area, New South Wales, Australia, a swarm of hydrothermal calcite and quartz veins is hosted in upright, open to close folded limestones and shales. Overprinting relationships and vein geometries demonstrate that the vein swarm formed progressively during fold growth and associated reverse faulting. Textures preserved in veins reveal that veins formed via hundreds to thousands of individual dilation and mineral precipitation events. Bedding-parallel flexural slip during fold growth was associated with laminated vein development, and limb-parallel stretching during fold growth was associated with the formation of bedding-orthogonal extension veins. The presence of subhorizontal extension fractures and severely misoriented reverse faults imply that fluid pressures exceeded lithostatic levels, at least transiently, during the development of the vein swarm. Vein δ18O compositions increase upwards through the Murrumbidgee Group in response to a progressive reaction of an externally derived, upwards-flowing low-δ18O fluid (of probable meteoric origin) with host limestones. Vein δ18O and 87Sr/86Sr compositions vary spatially and temporally within the same outcrop, and within individual veins. These variations are inferred to be caused by the ascent of packages of fluid along constantly changing flow pathways caused by multiple permeability creation–destruction cycles associated with fault slip and fault sealing. Vein trace and rare earth element (REE) concentrations are more variable, probably reflecting rapid rock buffering along fluid pathways on length scales of less than 10 m. Our results indicate that fluid-flow pathways change dynamically during crustal shortening, with pathways switching between states of low and high permeability during episodic fault slip and associated fracture development. Supplementary material: Two appendices are available at http://www.geolsoc.org.uk/SUP18492.

Syntectonic veins record information on spatial and temporal variations in the composition of fluids migrating through the crust during progressive deformation and associated vein growth (e.g. Dietrich et al. 1983;Rye & Bradbury 1988). Exhumed vein systems potentially contain a record of where fluid flow was localized, how flow has influenced the mechanical behaviour of the crust, and the nature of chemical reactions between rocks and fluids along fluid-flow pathways. Vein chemistry may also provide constraints on the nature of fluid reservoirs contributing to fluid flow during crustal deformation.
An extensive fold-and fault-related vein network (hereafter called the 'Taemas Vein Swarm' or TVS) composed mainly of calcite veins is developed throughout much of an interbedded limestone and shale sequence (the Murrumbidgee Group) in the Taemas area of southeastern NSW. The vein swarm is localized mainly within an area of approximately 20 km 2 . Vein development and folding were synchronous, with a variety of beddingdiscordant fault veins, bedding-parallel fault veins and extension veins developed. Laminated veins related to bedding-parallel slip are common, particularly in thinly interbedded limestone-mudstone units. Bedding-discordant faults, fault-related extension veins and extension veins related to flexural flow and bedding-parallel stretching of fold limbs are also found (Cox 2007).
In addition to isotopic compositions of calcite, we present trace element compositions from hydrothermal calcite veins. Calcite commonly contains appreciable concentrations of Mg, Mn and Fe, and lesser amounts of Sr and Ba (Deer et al. 1962(Deer et al. , 1992Mucci & Morse 1990). In general, there are little available experimental data addressing the trace element composition of calcite precipitated at temperatures above 50 8C. The concentration of trace elements in calcite will reflect a variety of factors in the fluid from which the calcite precipitated. These factors include the concentration of substituting elements in solution, temperature, pH, other aspects of fluid composition (e.g. concentrations of other species such as CO 3 22 ) and oxidation state. The major factors controlling trace element partitioning between calcite and fluid are the size of the cation site and the ionic radius of the substituting element (Garrels & Christ 1982;Mucci & Morse 1990;Zhong & Mucci 1995). The reviews of Mucci & Morse (1983) and Wasylenki et al. (2005) found a strong temperature control on the incorporation of both Mg and Sr into calcite. Faster precipitation rates also lead to higher concentrations of both Sr and Mg in calcite (Mucci & Morse 1990). Fe and Mn concentrations in calcite are controlled by the concentration of Fe and Mn in solution, as well as the temperature and precipitation rate of calcite (Dromgoole & Walter 1990). In hydrothermal fluids, rare earth element (REE) fractionation is a function of: (a) sorption and desorption of REEs during migration of fluids along particle surfaces; and (b) coprecipitation (Bau & Moller 1992).
The aim of this chapter is to use spatial and temporal variations in vein isotopic and trace element composition to explore controls on fracturing, fluid flow, and fluid -rock reaction, in a fracturecontrolled hydrothermal system that was active during progressive crustal shortening.

Geological setting of the Taemas Vein Swarm
The Taemas area is located SW of Yass, in the Eastern Belt of the Lachlan Orogen, in southeastern New South Wales, Australia (Glen 1992). Here, a major fold triplet (Wee Jasper Syncline, Narrangullen Anticline and Taemas Synclinorium) forms the larger Black Range Synclinorium (c. 180 km long Â 25 km wide). The Taemas Synclinorium is a doubly-plunging synclinorial structure approximately 5 km wide (Cramsie et al. 1975) (Fig. 1). Within the Taemas Synclinorium, sedimentary rocks of the Murrumbidgee Group (Table 1) have been folded into upright, open to close folds (wavelengths varying from tens-of-metres-to kilometrescale), which have a predominant NNW trend, and steep axial surfaces (Fig. 2). The synclinorium is to the west of the major, steeply-dipping, Warroo and Deakin-Devil's Pass Fault System. Here, Silurian volcanics overthrust the Murrumbidgee Group sediments (Cramsie et al. 1975). Within the Taemas Synclinorium, the Murrumbidgee Group is approximately 1 km thick, and is composed of several limestone and shale formations that were deposited during the Early Devonian (Browne 1958) (see Table 1 for stratigraphy and formation name abbreviations).
At the base of the sedimentary sequence (in the Sugarloaf Creek Formation, SLC and the underlying volcanic basement) bedding dips are low (typically ,208) and bedding is gently folded, with fold wavelengths of the order of hundreds of metres to more than 1 km. Higher in the sequence, particularly in the Cavan Bluff Limestone (CBL), Spirifer yassensis Limestone (SYL) and Bloomfield Limestone (BFL), close to tight folds occur, with shorter fold wavelengths (from tens to hundreds of metres: see Figs 2 & 3). The shorter fold wavelength in the interbedded limestone -shale units (SYL and BFL), compared with the thicker, more competent underlying and overlying units (Majurgong Formation (MJF) and Receptaculites Limestone, RCL) reflects a strong control of thickness of competent stratigraphical units on fold wavelength. The changes in fold wavelength and strains accommodated by folding through the Murrumbidgee Group require detachment accommodated by faulting where major changes in fold wavelength occur.
Cleavage is strongly developed in the MJF, and in shale-rich beds in the SYL and BFL. Within sandstone beds in the MJF, cleavage fans strongly around outcrop-scale folds, and is typically at a high angle to bedding. This implies that layerparallel shortening occurred early during fold growth, with the subsequent buckling of bedding (Ramsay & Huber 1987). In general, shales have a penetrative slaty cleavage, whereas limestone beds are internally relatively unstrained or contain a weak pressure-solution cleavage. Total strain, calculated using bed lengths of internally unstrained limestone sequences, is in the range of 20-50%. Given geological strain rates of the order of 10 214 -10 215 s 21 (Pfiffner & Ramsay 1982;  Mueller et al. 2000), this would imply that folding occurred over a period of 0.5-15 Ma.

Bedding-parallel veins
Bedding-parallel veins commonly contain tens to hundreds of mesoscopic grey-brown laminations, which lie subparallel to the vein margins. Laminations are usually striated, with striations generally subperpendicular to the plunge of fold hinges. However, the trend of striations on low-angle bedding-parallel veins at Shark's Mouth Peninsula (Fig. 2, locality 4) varied over 558 on different laminae in the same vein. Most veins have dips of 408-608. The laminated, bedding-parallel veins are analogous to those described previously, where slickenfibres and slickenlines found on laminations in the bedding-parallel veins record the slip vector (Gaviglio 1986;Tanner 1989;Jessell et al. 1994;Fowler 1996 and references therein). Some BPV may be traced around fold hinges. The presence of BPV in fold-hinge zones, and asymmetrically folded laminations within BPV on fold limbs, suggest that ongoing fold growth post-dated initial flexural slip. It seems likely that during the initial stages of folding, significant strain was accommodated via slip along bedding. However, as bedding dips increased during folding, frictional lock-up occurred (Ramsay 1974). During ongoing fold tightening, new bedding-discordant faults have formed. The occurrence of small saddle reefs indicates that dilation occurred at some fold hinges during fold amplification by flexural slip.

Extension veins
Fold limbs rotated into steep orientations have undergone limb-parallel stretching, resulting in incipient bedding boudinage and formation of associated subhorizontal extension veins (Fig. 5). Extension veining of this type is most prevalent in the CJL. This is likely owing to the high competence of this unit relative to the surrounding interbedded limestone -shale of the SYL and BFL.
Generally, extension veins overprint cleavage development and are generally later than beddingparallel slip veins. At locality 3 in Figure 1, calcite extension veins cut cleavage at a high angle on fold limbs. A bedding-parallel vein between the two anticlines is cut by extension veins, which dip east (3428/378E). The bedding-parallel vein is interpreted to be the result of flexural slip during folding. The high angle of calcite extension veins to cleavage within folds is consistent with some veins formed as beds deformed via flexural flow (Ramsay & Huber 1987). Extension veins crosscutting the bedding-parallel vein imply that strain may be accommodated via flexural-flow folding after bedding-parallel slip ceases due to the frictional lock-up of beds (Ramsay 1974). It is noted that en echelon arrays of veins related to flexural flow form only in semi-competent and incompetent beds, and do not form within more competent massive limestone beds.
Extension veins can also show mutually overprinting relationships with cleavage (in shale-rich beds) and stylolites (in more massive limestones), and folded veins are rare. This implies that extension vein formation and folding were contemporaneous. Extension veins sometimes show mutually overprinting relationships (Fig. 5c, e), implying that the orientation of s 3 (at least sometimes) changed dynamically over time.
In summary, various vein types formed throughout crustal shortening. Thus, vein chemistry may be used to track variations in fluid composition  Fig. 1). Note the decoupling that must occur between and within SYL and CJL limestones to allow observed fold geometries. Black rectangles mark the locations of the two outcrops shown in Figure 12 and the outcrop shown in Figure 5 (as indicated by numbers in parentheses in the figure).
(d 18 O, 87 Sr/ 86 Sr, trace and rare earth elements), not just during growth of individual veins but also as the TVS evolved from early fold growth to later fold tightening, cleavage development and associated reverse faulting.

Vein textures
Veins dominantly have fibrous (Fig. 6a), massive (Fig. 6b), laminated (Fig. 6c) and fibrous ( Fig. 6d) textures, with elongate-blocky and crustiform textures also occurring (Bons 2000;Oliver & Bons 2001). Bedding-parallel and discordant fault veins have massive and laminated textures, with fibrous textures preserved in some parts of fault and beddingparallel veins (Fig. 6b). Conversely, extension veins have massive, elongate-blocky or fibrous textures (Fig. 6a, d). It is emphasized here that the particular textures are not isolated to specific host lithologies, and textural variations are present among veins within the same outcrop, as well as within individual veins (Fig. 6b). Notably, veins may contain both fibrous and massive calcite; massive and fibrous extension veins also show mutually overprinting relations (Fig. 6b).
Laminated textures in bedding-parallel slip veins have been classified according to the terminology of Koehn & Passchier (2000). Inclusion bands are thin, dark bands parallel to the vein margins. Crack -seal bands are thin, dark bands between parallel inclusion bands, which, in this study, are typically inclined at angles of 208 -358 to inclusion bands (Figs 6c & 7). Crack -seal bands are typically separated by distances of 100 mm-2 mm in laminated veins (Fig. 6c), and hundreds of crack -seal bands may occur in an interval of around 10 cm along one calcite lamina (Fig. 6c). Crack -seal bands are inferred to have formed along dilational

Sample collection
The position and structural relationships of veins were recorded in the field prior to sample collection (see Supplementary material SUP18492 for sampling locations and complete analytical results). Unaltered host rock (at distances of more than 50 m from any visible veins) was collected from the Cavan Bluff, Spirifer yassensis, Currajong, Bloomfield and Receptaculites limestones.

Stable isotope analysis
Oxygen and carbon isotope ratios for carbonates were measured on a Finnigan MAT251 mass    (Coplen et al. 1983).
Isotope results have been normalized on the VSMOW and VPDB scales so that analyses of: The standard deviation (2s) for the 50 replicate NBS-19 standards used during the analysis of these samples was 0.02‰ for d 13 C and 0.06‰ for d 18 O.
Quartz oxygen isotope ratios were measured at the University of New Mexico. Quartz samples were separated from calcite (the calcite was stored for oxygen isotope analysis, as outlined earlier), and the quartz chips were placed into a 1 M HCl solution until all effervescence had ceased and calcite was removed.
The resulting quartz separates were examined using a binocular microscope, and clear quartz pieces were selected for analysis by laser fluorination (following the method of Sharp 1990). An internal laboratory standard (Lausanne-1) was analysed to determine the reproducibility of analyses (+0.2‰).

Strontium isotope analyses
Host rock. Sr isotope compositions of host-rock carbonate were measured by thermal ionizing mass spectrometry (TIMS). Approximately 1 kg of each of two samples from each limestone member (10 samples in total) was crushed using a tungsten carbide swing mill. Around 0.06-0.11 g of rock powder was placed into a clean Teflon w screw-cap vial. To separate only the carbonate component of the limestones, 1 ml of distilled 1 M acetic acid was added to each beaker, resulting in immediate, gentle effervescence. The beakers were allowed to rest for 2 h at room temperature, and a further 1 ml of acetic acid was added, after which effervescence ceased. The resulting mixture of solution and sediment was centrifuged in clean tubes, and the liquid was drawn off and dried on hot plates. The samples were then taken up in HNO 3 for loading onto cation-exchange columns, where Sr was separated from other matrix elements. Rubidium was not collected or analysed as Rb concentrations in carbonate are generally very low (Faure & Powell 1972), and were confirmed to be very low (,1 ppm) by laser ablation multi-collector inductively coupled plasma mass spectrometry (LA-MC-ICP-MS) analyses (see later).
Purified Sr was loaded in H 3 PO 4 on Ta filaments and analysed on a Finnigan Mat261 mass spectrometer. All filaments were out-gassed for 30 min prior to loading. Sr isotope values were normalized to 86 Sr/ 88 Sr ¼ 0.1194. The NIST SRM-987 Sr isotope standard analysed had a 87 Sr/ 86 Sr value of 0.71023 + 0.00001.
Carbonate veins. Sr isotope compositions of vein calcite were analysed on the same samples on which trace element analyses were conducted. Analyses were made using in situ LA-MC-ICP-MS, and the results of the vein analyses are reported in Supplementary material SUP18492. Analyses were carried out using a HelEx ArF excimer laser ablation system, interfaced to a Finnigan MAT Neptune MC-ICP-MS (see Eggins et al. 1998Eggins et al. , 2005 for details). Analyses were performed using a singlespot approach (with a laser spot diameter of 137-233 mm). Laser pulse rates of 5 Hz in combination with a laser fluence of 5 J cm 21 , short laser wavelength (l ¼ 193 nm) and aperture imaging optics were used to attain controlled calcite ablation (c. 1 mm s 21 ) in a He ablation medium (Eggins et al. 1998).
The gas flow and electrostatic lens settings were optimized for maximum Sr sensitivity and peak shape while ablating a modern Tridachna clam shell, which has a measured 87 Sr/ 86 Sr value of 0.709143 + 15 (Woodhead et al. 2005). Tridachna was additionally used to monitor instrument reproducibility and accuracy. For 22 analyses of Tridachna, the average 87 Sr/ 86 Sr ratio was 0.709149 + 38 (2s). Further details of analytical procedures are given in Barker et al. (2006).

Trace element analyses
Host rock. The trace element compositions of the carbonate component of limestone host rocks were measured by solution collector inductively coupled plasma mass spectrometry (ICP-MS). A known amount of approximately 50-60 mg of rock powder was placed into a clean Teflon w screw-cap vial. To separate only the carbonate component of the limestones, 1 ml of distilled 1 M acetic acid was added to each beaker resulting in immediate, gentle effervescence. The beakers were allowed to rest for 2 h at room temperature, and a further 1 ml of acetic acid was added, after which effervescence ceased. The resulting mixture of solution and sediment was centrifuged in clean tubes, and the liquid was drawn off, dried on hot plates and taken up in around 100 ml of 2% HNO 3 for analysis. Trace elements were measured using a quadrupole ICP-MS (Agilent 7500s) running in solution mode. Multiple major and trace elements ( 9 Be Lu, 185 Re, 208 Pb, 209 Bi, 232 Th, 238 U) were analysed, with multiple internal standards (Be, As, In, Re, Bi) to monitor instrument performance and drift.
Carbonate veins. Veins were analysed for trace element compositions using a multiple-spot analysis, laser ablation ICP-MS approach. Samples were analysed using a pulsed ArF Excimer laser (l ¼ 193 nm) and a quadrupole ICP-MS (Agilent 7500s; Eggins et al. 1998). Samples were precleaned with ethanol. Every vein sample was analysed between three and 10 times using a 70 mm laser spot, and an average composition was determined (results reported in Supplementary material SUP18492). This approach has some drawbacks, particularly regarding zoning of trace elements within individual crystals in carbonate veins. However, this approach was chosen to: (a) avoid inclusions within veins; (b) minimize sample preparation time; and (c) allow veins with complex internal fabrics to be documented more completely. Multiple major and trace elements ( 23 Na, 24  were simultaneously analysed during laser sampling by repeated, rapid sequential peak hopping, with a mass spectrometer cycle time of 0.65 s. Data reduction followed established protocols for timeresolved analysis (Longerich et al. 1996), with NIST 612 standard (values of Pearce et al. 1997), analysed before and after every six samples, and 43 Ca was used as an internal standard. Laser pulse rates of 5 Hz in combination with a laser fluence of 5 J cm 22 , short laser wavelength (l ¼ 193 nm) and aperture imaging optics were used to attain controlled calcite ablation (c. 1 mm s 21 ) in a He ablation medium (Eggins et al. 1998).

Vein and wall-rock chemistry
Characterizing the isotopic and trace element composition of a heterogeneous host rock is a challenging problem. For example, it is difficult to determine the overall oxygen or strontium isotope composition of a calcareous shale because small variations in the proportion of carbonate:silicate minerals may cause significant changes in Sr or O isotope ratios. In addition, it is not feasible to characterize each individual bed in an interbedded limestone -shale unit and then produce an 'average' isotopic composition. The majority of veins in the Taemas Vein Swarm are calcite, with minor quartz. Cox (2007) found that the siliciclastic Majurgong Formation had no affect on the oxygen isotope composition of fluids migrating through this unit. Therefore, it was decided to characterize only the carbonate component of host rocks via weak acid extraction of carbonate. In particular, extracting only the Rb-poor carbonate component of host rocks means that no age correction is required to determine 87 Sr/ 86 Sr values at the time of hydrothermal vein growth (Faure & Powell 1972).
Distal limestone samples (more than 50 m from significant veining) have d 18 O between 23 and 25‰, d 13 C of 20.5 to þ3.0‰ (Cox 2007), and 87 Sr/ 86 Sr ratios between 0.70815 and 0.70828 (Table 2), which are consistent with sea-water values for the Early Devonian (Veizer et al. 1999). Host rocks measured within about 1 mm-1 cm of vein margins have a broader range of d 18 O, varying between 14 and 24‰ (Cox 2007). Depleted d 18 O margins are usually a few centimetres wide, except around a fault zone at the bottom of the Cavan Bluff Limestone (Fig. 4), where d 18 Odepleted zones are several metres wide (Cox 2007 (Fig. 4). Oxygen isotope compositions from calcite and quartz inferred to have grown in equilibrium (i.e. intergrown fibrous quartz and calcite) suggest fluid temperatures of between 100 and 250 8C (Table 3 and Fig. 9) (thermometer of Zheng 1993).
Host-rock carbonates have variable Mg concentrations, particularly in the Cavan Bluff Limestone (see Table 4    In the interbedded limestone -shale lithologies, veins have the most variable trace element concentrations, particularly in the Cavan Bluff Limestone (Fig. 11)

Chemical variations within and between outcrops
Veins were assessed on an outcrop-by-outcrop, as well as within-outcrop, scale to examine how fluid-rock reaction varies between veins in a similar host rock. Figure 4 shows Figure 12a, b illustrates the structural setting, and Sr and O isotope ratios, of veins from distinct but closely spaced outcrops of folded Spirifer yassensis Limestone at Kangaroo Flat (locality 1 in Fig. 1, which is also shown in Fig. 3). These outcrops represent one of the few locations in which the chemistry of veins from the same stratigraphical level, but different structural settings, may be compared (see highlighted regions in Fig. 3). Site A (Fig. 12a) has a bedding-parallel vein developed within an asymmetrical syncline -anticline pair.
Similar trends are observed for trace elements, with veins in the SYL having a broad range of trace element compositions, with no relationship observed between vein type and trace element concentration (Fig. 15). In comparison, extension veins within the CJL have lower concentrations of most  Table 3), and carbonate veins (black circles) with stratigraphical height through the Murrumbidgee Group (see Supplementary Appendix 2). Limestone formations and members are the same as in Figure 10.
trace elements compared to fault or bedding-parallel veins (Fig. 16). In particular, REE concentrations are lower in extension veins than in bedding-parallel or fault veins. Extension veins in the Currajong Limestone have Sr isotope and trace element compositions that are most similar to host-rock carbonate compositions. In comparison, all veins in the SYL have a broad range of Sr isotope ratios and trace element concentrations, which are generally elevated above host-rock carbonate compositions. Many extension veins in the CJL are related to fold limb-parallel stretching. In comparison, most extension veins in the Spirifer yassensis Limestone are associated with bedding-discordant fault zones. The fluid that migrated through late extension veins in the CJL was apparently significantly more influenced by local host rock than the fluid from which fault veins or bedding-parallel veins formed.

Discussion
The isotopic and trace element composition of hydrothermal calcite veins found within fault-fracture systems will reflect a variety of factors, including changes in fluid source, fluid-flow pathways, and the intensity of fluid -rock reaction in both time and space. The veins of the Taemas Vein Swarm record spatial and temporal variations in the composition of hydrothermal fluids responsible for the formation of the vein swarm.

Fluid sources
If it is assumed that calcite precipitated in equilibrium with the parent fluid, and the fluid temperature can be estimated, then the d 18 O of vein calcite may be used to determine the isotopic composition of the parent fluid. Quartz -calcite oxygen isotope pairs (see Fig. 10) (Cox 2007) (Sheppard 1986). To form a meteoric fluid with these low d 18 O values, the meteoric fluid must have been sourced at either high latitude (c. 50 -608), or significant altitude (2-4 km) (Bowen & Wilkinson 2002). Several veins, or sections, of individual veins have unusual 87 Sr/ 86 Sr and d 13 C values compared to the majority of veins in the TVS (e.g. sample C2a; see Supplementary material SUP18492). Barker et al. (2006) and Cox (2007) attributed localized negative-d 13 C values to the oxidation of organic matter during fluid -rock reaction. The coupling of elevated 87 Sr/ 86 Sr and depleted d 18 O values implies that a second fluid 'source' may have been present. This fluid may have been a residual interstitial pore fluid, which underwent enhanced fluidrock reaction with iron oxides (causing oxidation of organic carbon) and 87 Sr-enriched clay minerals (i.e. Majurgong Formation or the SYL) prior to mixing with the invading meteoric fluid.

Implications of isotopic and trace element data for fluid flow
Assuming that calcite deposition in veins was an equilibrium process, then the d 18 O of vein carbonate changes in response to the d 18 O of pore fluid, as pore fluid and wall rock undergo progressive reaction (Fig. 17). The conceptual model presented in Figure 17 emphasizes that veins within the same outcrop could have considerably different isotopic compositions (compare vein pair 1 and 3, and vein pair 6 and 7 in Fig. 4), depending on the length of the pathways along which those fluids flow. If a fluid migrates along a tortuous (i.e. long) fluid-flow pathway, then it will have greater opportunity to react with the country rock than fluids that migrate along more direct (i.e. shorter) fluid-flow pathways (see Fig. 17). In addition, fluid-flow pathways may change during deformation as permeability is created and destroyed. Variations in the isotopic composition of different vein types within the same outcrop in the Currajong Limestone indicate that fluids that precipitated calcite in fault and bedding-parallel veins underwent less interaction with host rocks than fluids forming extension veins. This is probably related to the length of fluidflow pathways, with fault and bedding-parallel veins having (relatively) short flow pathways, causing fluids to undergo less reaction with host rock than fluids that formed extension veins (see Figs 16 & 17). This suggests that bedding-parallel and fault veins act as high-permeability pathways along which fluid could migrate.
The transition from the SYL to CJL is marked by a change from mixed shale -limestone lithology to only limestone, and a coincident decrease in vein 87 Sr/ 86 Sr ratio. The carbonate content of SYL limestone determined in the laboratory (estimated from loss of weight after acetic acid leaching) is  as little as about 10 m stratigraphically above the Currajong Limestone. This implies that fluids have scavenged trace elements from the surface of other minerals such as layer silicates, and/or feldspars, over relatively short reactive path lengths during fluid flow. The lack of mineral specific Sr isotope data, and the uncertainty regarding the timing of hydrothermal activity (see Barker et al. 2009), make this hypothesis difficult to test further. During this study a relatively small number of host rocks were analysed. In particular, analyses of shales were not carried out during this study. The consistent isotopic and trace element composition of limestone carbonate throughout the stratigraphical sequence suggests that average carbonate compositions do not vary by a significant degree (except, perhaps, for Mg concentrations), particularly compared to the degree of variation in trace element concentrations in veins. The study of Cox (2007) demonstrated that fluids dominantly   (Lassey & Blattner 1988;Moller et al. 1991;Bau & Moller 1992;DePaolo & Getty 1996;Steefel & Lichtner 1998;Hecht et al. 1999;DePaolo 2006): † spatial and temporal variations in the composition of infiltrating (source) fluid; † interaction of infiltrating fluid with matrix fluid; † dissolution of host-rock minerals, releasing lattice-bound trace and minor elements; that is, host carbonate being dissolved; † scavenging by fluids of trace elements loosely bound to mineral surfaces, particularly layer silicates, releasing adsorbed elements. In addition, shales generally have well-developed cleavages, indicating that layer silicates have deformed and recrystallized, implying that chemical species associated with these minerals would be available for infiltrating fluids during deformation; † precipitation of hydrothermal minerals (e.g. calcite) leading to coprecipitation of other trace elements, removing those elements from solution; † sorption of trace elements onto mineral surfaces during fluid migration, removing trace elements from fluids; † changes in physicochemical conditions (e.g. temperature, pressure, pH, complexing species), which may affect trace element fractionation fluids and vein-forming minerals.  Lichtner (1998) andDePaolo (2006) provided models addressing the chemical interaction of host rock, 'matrix fluid' and 'fracture fluid' during fracture-controlled fluid flow. As fluid migrates through a fracture, diffusion of trace elements occurs between that fluid and the rock matrix. The position of alteration fronts will be dependent on the temperature, fluid-flow rate, diffusion rate, kinetics of reactions and the reactive surface area of minerals. Along any one fracture, these factors will change depending on the fracture aperture, fracture roughness and wall-rock composition. For example, a small cataclastic fault zone, containing pulverized host rock with a high reactive surface area, will react in a different manner to a smoothsided extension fracture through the same host rock.
Trace element concentrations are highest in veins in the lowest stratigraphical units; that is, those contained within the Cavan Bluff Limestone. This implies that the fluid which infiltrated the base of the Murrumbidgee Group was already enriched in trace elements. Presumably, the trace element composition of fluids infiltrating the base of the CBF reflects fluid -rock interaction that occurred with underlying rocks (i.e. volcanic sediments of the Black Range Group). Notable in some veins within the Cavan Bluff Limestone are elevated Sr concentrations, and veins with extremely positive Eu anomalies that could be produced by the dissolution of Ca-feldspar, which has appreciable concentrations of both Sr and Eu (Schnetzler & Philpotts 1970). When vein REE concentrations are normalized to chondrite values, it is apparent that REE patterns in Cavan Bluff veins are generally LREE-enriched, whereas veins higher in stratigraphy in the SYL and CJL become gradually depleted in LREE (Figs 11,15 & 16). In hydrothermal fluids, REE fractionation is a function of: (a) sorption and desorption of REEs during the migration of fluids along particle surfaces; and (b) co-precipitation (Bau & Moller 1992). Carbonate and hydroxyl ligands form stronger complexes with HREEs than LREEs (Bau & Moller 1992  calcite precipitating from a solution with higher CO 3 22 concentrations, which will be relatively enriched in the LREEs. This is because in carbonate-poor solutions there will be little difference in the complexing of the light and heavy REEs (Bau 1991;Bau & Moller 1992). If this interpretation is correct, it is suggested that [CO 3 22 ] decreased as fluids migrated upwards through the Murrumbidgee Group, consistent with decreasing fluid pressure as fluids ascended.
Implications for fluid-flow pathways and vein development during crustal shortening Important conclusions that may be drawn from this study, and the previous studies of the Taemas vein swarm (Barker et al. 2006;Cox 2007;Barker et al. 2009) Barker et al. (2006) and Cox (2007), reveal that fluid-flow pathways, path lengths, and/ or fluid-flow and fluid -rock reaction rates changed dynamically during the growth of individual veins, and between different veins in the same outcrop. This is probably related to the creation and destruction of permeability during repeated fracture opening and sealing events. Dynamic switches in fluid-flow pathways are interpreted in terms of fluid flow through a fault -fracture mesh (Hill 1977;Sibson 2001). In such a mesh, it is predicted that fractures will be transiently permeable after fracturing (promoting rapid migration of fluids) and then this permeability will be destroyed (i.e. via hydrothermal mineral deposition), thus creating a dynamic fluid-flow environment.
According to the conditions for tensile failure, fluid pressure must exceed the least compressive stress and the tensile strength of the rock to cause extension fractures to form. In a contractional (i.e. reverse) faulting regime, gently dipping extension veins imply that: (a) the least compressive stress was approximately vertical; and (b) that the fluid pressure (at least transiently) exceeded the lithostatic pressure (i.e. l v . 1.0: Sibson 2001). Additional evidence for transiently high fluid pressures is provided by steeply dipping bedding-parallel slip veins, and severely misoriented beddingdiscordant reverse faults. The presence of folded and unfolded laminations in some of these veins suggests that slip continued on these veins throughout fold growth, with some vein dips exceeding 708. For slip to continue at severe misorientation on incohesive reverse faults requires that fluid pressures exceed supralithostatic levels (Sibson 1985). This constraint must also be met for cohesive faults (Cox 2010). These factors indicate that the formation of the Taemas Vein Swarm was driven by overpressured fluids (cf. Cox 2007).
Differential stress levels must have varied significantly during deformation (at least on a local scale). Parts of the stratigraphy must have had low differential stress levels at the time of vein formation to form extension fractures. Mutually overprinting relationships between shear and extension fractures indicate that differential stress values oscillated between (s 1 2 s 3 ) , 4T and (s 1 2 s 3 ) . 5.66T, where T is the tensile strength of the rock (Secor 1965;Hancock 1985). Such differential stress variations could be related to the loading and stress release associated with repeated seismic slip events (Sibson 1989).
In active fold-and-thrust belts, strain is accommodated in sedimentary rocks by a combination of folding and thrust faulting (e.g. Shaw & Suppe 1994). Seismic reflection profiles and surface mapping suggest that actively growing folds are intimately related to seismically active faults (e.g. Namson & Davis 1988;Davis et al. 1989;Shaw & Suppe 1994). In the Taemas Vein Swarm, timing relationships between bedding-parallel slip veins and overprinting flexural flow and extension veins indicate that early during folding strain was accommodated via flexural-slip folding with associated fluid flow along bedding surfaces. Crack -seal textures in bedding-parallel veins at Taemas indicate that bedding-parallel slip was episodic, suggesting that fold growth occurred episodically. Episodic fold growth and associated faulting probably generate significant fracture permeability in actively deforming fold-thrust belts (Finkbeiner et al. 1997). Later in folding, once fold lock-up occurred, fluid flow may have localized more along bedding-discordant faults and extension vein networks associated with fold limb stretching, rather than being isolated along bedding planes. Where bedding-parallel veins linked with discordant fault veins (i.e. thrust faults cutting through fold hinges), it is likely that some bedding-parallel veins remained active as faults, even though fold tightening had essentially ceased.

Conclusions
The Taemas Vein Swarm preserves evidence of hydrothermal vein growth during fold growth. Vein formation was intimately related to space created during folding, with vein growth active from the early to the latest stages of fold growth. Mutually cross-cutting relationships between veins indicate that vein growth was intermittent, and that veins formed as localized stress fields underwent significant changes in both orientation and magnitude.
Subhorizontal extension fractures and severely misoriented faults indicate that fluid pressures intermittently exceeded lithostatic levels over (at least) local regions of the deforming crust. Individual veins grew incrementally and preserve a variety of textures (fibrous, massive and laminated), and indicate that vein opening and mineral deposition rates varied significantly. Mutually cross-cutting veins and incrementally developed vein textures imply that high-permeability fluid-flow pathways varied dynamically through time and space.
A progressive increase in vein d 18 O with increasing height through the Murrumbidgee Group is attributed to progressive buffering of 18 O-depleted fluids by reaction with host-rock carbonate. Vein d 18 O values imply that the invading fluid had d 18 O compositions consistent with a meteoric fluid source. The isotopic and trace element composition of hydrothermal veins is variable across stratigraphy, between outcrops, within individual outcrops and within individual hydrothermal veins. Variability in O and Sr isotope ratios within and between veins is attributed to variable fluid -rock reaction along dynamically changing fluid-flow pathways, caused by episodic failure, permeability enhancement and transitory fluid flow accompanying episodic slip events within a fault-fracture mesh.
The isotopic and trace element composition of syntectonic veins reflect different rates of fluidrock reaction for different elements and isotopic systems during hydrothermal fluid flow. Strontium isotope ratios ( 87 Sr/ 86 Sr) indicate that fluids react with host-rock carbonate, but also scavenge Sr from Sr-enriched minerals (e.g. illite), within shale beds over short reactive path lengths. Trace element concentrations generally decrease with increasing stratigraphical height. Rare earth element concentrations and patterns are influenced by progressive calcite precipitation and sorption along fluid-flow pathways, and changes in REE complexation in solution.
Thanks to K. and S. Kilpatrick for land access and their friendship. Thanks to M. Gagan for facilitating access to stable isotope analyses, and J. Cali and H. Scott-Gagan for providing able technical help with stable isotope mass spectrometry. S. Eggins and L. Kinsley provided assistance with Sr isotope analyses. Thanks to C. Allen, M. Shelley and C. Magee for assistance with trace element LA-ICP-MS, and M. Norman for help with solution ICP-MS analysis. The Mervyn and Katalin Paterson Fellowship allowed S. L. L. Barker to travel to UNM to carry out analytical work. Thanks to Z. Sharp and V. Atudorei for assistance with quartz stable isotope analysis, and V. Atudorei for hospitality. S. L. L. Barker acknowledges receipt of an Australian Postgraduate Award, and RSES and the ANU for additional scholarship support during his PhD studies. S. F. Cox acknowledges ARC grant DP 0452448. An anonymous reviewer and J. Urai are thanked for helpful reviews, and Å . Fagereng is thanked for editorial guidance.