It is surprising to encounter active saline spring activity at a constant 6°C temperature year-round not far away from the North Pole, at latitude 79°24N, where the permafrost is ca. 600 m thick and average annual temperature is -15°C. These perennial springs in Expedition Fiord, Queen Elizabeth Islands, Canadian Arctic Archipelago, had previously been explained as a recent, periglacial process. However, the discovery near White Glacier (79°26.66N; 90°42.20W; 350 m.a.s.l.) of a network of veins of hydrothermal origin with a similar mineralogy to travertine precipitates formed by the springs suggests that their fluids have much deeper circulation and are related to evaporite structures (salt diapirs) that underlie the area. The relatively high minimum trapping temperature of the fluid inclusions (avg. ~200 ± 45°C, 1σ) in carbonate and quartz in the vein array, and in quartz veins west of the site, explains a local thermal anomaly detected through low-temperature thermochronology. This paper reviews and updates descriptive features of the perennial springs in Expedition Fiord and compares their mineralogy, geochemistry, and geology to the vein array by White Glacier, which is interpreted as a hydrothermal predecessor of the springs. The perennial springs in Axel Heiberg Island are known for half a century and have been extensively described in the literature. Discharging spring waters are hypersaline (1-4 molal NaCl; ~5 to 19 wt% NaCl) and precipitate Fe-sulfides, sulfates, carbonates, and halides with acicular and banded textures representing discharge pulsations. At several sites, waters and sediments by spring outlets host microbial communities that are supported by carbon- and energy-rich reduced substrates including sulfur and methane. They have been studied as possible analogs for life-supporting environments in Mars. The vein array at White Glacier consists of steep to subhorizontal veins, mineralized fractures, and breccias within a gossan area of ca. 350 × 50 m. The host rock is altered diabase and a chaotic matrix-supported breccia composed of limestone, sandstone, and anhydrite-gypsum. Mineralization consists of brown calcite (pseudomorph after aragonite) in radial aggregates as linings of fractures and cavities, with transparent, sparry calcite and quartz at the centre of larger cavities. Abundant sulfides pyrite and marcasite and minor chalcopyrite, sphalerite, and galena occur in masses and veins, much like in base metal deposits known as Mississippi Valley Type; their weathering is responsible for brown Fe oxides forming a gossan. Epidote and chlorite rim veins where the host rock is Fe- and Mg-rich diabase. The banded carbonate textures with organic matter and sulfides are reminiscent of textures observed in mineral precipitates forming in the active springs at Colour Peak Diapir. Very small fluid inclusions (5-10 μm) in two generations of vein calcite (hexagonal, early brown calcite we denominate “cal1” lining vein walls; white-orange sparry calcite “cal2” infilling veins) have bulk salinities that transition between an early, high-salinity end-member brine (up to ~20 wt% NaCl equivalent) to a later, low-salinity end-member fluid (nearly pure water) and show large fluctuations in salinity with time. Inclusions that occupy secondary planes and also growth zones in the later calcite infilling (deemed primary) have ranging from 100°C to 300°C (, ; independent of salinity), 2 orders of magnitude higher than average discharging water temperatures of 6°C at Colour Peak Diapir. Carbon isotope composition (δ13CVPDB) of the White Glacier vein array carbonates ranges from approximately -20 to -30‰, like carbonates formed by the degradation of petroleum, whereas carbonates at Colour Peak Diapir springs have a value of -10‰. Oxygen isotope composition (δ18OVSMOW) of vein carbonates ranges from -0.3‰ to +3.5‰, compatible with a coeval fluid at 250°C with a composition from -3.5‰ to -7.0‰. These data are consistent with carbonates having precipitated from mixtures of heated formational waters and high-latitude meteoric waters. In contrast, the δ18OVSMOW value for carbonates at Colour Peak Diapir springs is +10‰, derived from high-latitude meteoric waters at 6°C. The sulfur isotope (δ34SVCDT) composition of Fe-sulfides at the perennial springs is +19.2‰, similar to the δ34SVCDT of Carboniferous-age sulfate of the diapirs and consistent with low-temperature microbial reduction of finite (closed-system) sulfate. The δ34SVCDT values of Fe-sulfides in the vein array range from -2.7‰ to +16.4‰, possibly reflecting higher formation temperatures involving reduction of sulfate by organics. We suggest that the similar setting, mineralogical compositions, and textures between the hydrothermal vein array and the active Colour Peak Diapir springs imply a kinship. We suggest that overpressured basinal fluids expelled from the sedimentary package and deforming salt bodies at depth during regional compressional tectonic deformation ca. 50 million years ago (Eocene) during what is known as the Eurekan Orogeny created (by hydrofracturing) the vein array at White Glacier (and probably other similar ones), and the network of conduits created continued to be a pathway around salt bodies for deeply circulating fluids to this day. Fluid inclusion data suggest that the ancient conduit system was at one point too hot to support life but may have been since colonized by microorganisms as the system cooled. Thermochronology data suggest that the hydrologic system cooled to temperatures possibly sustaining life about 10 million years ago, making it since then a viable analogue environment for the establishment of microbial life in similar situations on other planets.

1. Introduction

The presence in a polar region of active saline springs at constant ca. 6°C temperature year-round, where average temperature is -15°C [1] and continuous permafrost is 600 m thick, is astonishing. This is the case in western Axel Heiberg Island (Figure 1), Queen Elizabeth Islands, Canadian Arctic Archipelago, at latitude 79°24N (e.g. [24]). The perennial springs (hereforth PSS) are situated in the periphery of evaporite diapirs (bodies of salt, anhydrite, and gypsum) that underlie the Expedition Fiord area (Figures 2 and 3(a)) and have risen over time as a ductile solid from their ultimate source in beds at ca. 10 km depth in the Sverdrup sedimentary basin (Figure 1; e.g., [5, 6]). The springs have been attributed to Quaternary periglacial processes (e.g., [7]). However, during a regional study of the uplift and thermal history of the archipelago using apatite fission-track thermochronology [810], it was noticed that the area where the PSS occur at the head of Expedition Fiord cooled to ca. 100°C between 20 and 30 million years later than the rest of the region (Figure 4). This realization led to a search for evidence for the pervasive circulation of warm fluids in the area in the form of veins with minerals amenable to fluid inclusion studies. In 2006, a network or array of mineral veins of hydrothermal origin was identified adjacent to White Glacier (Figures 2 and 3(b)). The White Glacier vein array (hereforth WGVA) is accompanied by a colourful gossan arising from the weathering of base metal (Fe, Cu, Zn, and Pb) sulfides and is the subject of this paper. We first update the descriptive features of the local PSS and then compare the WGVA with the prominent Colour Peak Diapir, ~13.7 km distant (Figures 2 and 3(a)), and two other PSS at Expedition Diapir and Wolf Diapir in terms of mineralogy, fluid inclusion systematics, and geochemistry, including stable O, C, and S isotope compositions. Based on the above studies and the literature, we conclude that the PSS have a kinship with the WGVA and the latter owes its existence to focussed expulsion of overpressured, hot, basinal fluids from an evaporite-dominated basement during tectonic compression known as the Eurekan Orogeny predominantly during the Eocene [11, 12]. The WGVA thus may represent the establishment of a plumbing system that has been later exploited by the PSS. The WGVA also is evidence for a regional metallogenic process that may have formed mineral concentrations elsewhere.

Permafrost at high latitudes is traditionally considered to effectively separate groundwater flow systems into two zones: a shallow, supra-permafrost zone that is active during summer months when air temperatures exceed 0°C and a sub-permafrost region that is hydrologically disconnected from the Earth’s surface [13]. The impervious nature of continuous permafrost means that such systems are poorly understood [14]; having a window into the subsurface through groundwater spring activity, therefore, provides a rare opportunity to further our understanding of hydrologic activity in polar environments, to explore life beneath permafrost, and to improve our appreciation of the potential for bioavailable water and subsurface life to exist on other planets such as Mars [1518]. The update of data on these anomalous polar springs and new interpretation of their process is a contribution to the understanding of geothermal systems in salt-dominated terrains.

2. Geological Setting

The perennial springs and WGVA lie within a relatively small area of the extensive (1000 km by 350 km) Sverdrup Basin (Figure 1), which contains a thickness of over 13 km of sedimentary and volcanic rocks ranging in age from Carboniferous to Tertiary [19]. The basin has been extensively explored for petroleum [20], although the wells in the eastern Sverdrup Basin have been disappointing and are more likely to produce gas than oil, due to the intervention of tectonism and the thermal effects of igneous intrusions [21]. The geology of the Expedition Fiord area (Figure 2) was initially mapped during the Jacobsen-McGill Arctic Research Expedition (1959-1962), among others by Tozer [22], Thorsteinsson and Tozer [23], and Fricker [24]. The evaporite diapirs were described by Hoen [25, 26]. A new map of the area including both Expedition Fiord and Strand Fiord was produced by Harrison and Jackson [27] and salt-induced structures reinterpreted by Jackson and Harrison [6]; their latter paper provides a valuable geological introduction to this study. The oldest rock unit exposed in the immediate area is allochthonous gypsum and anhydrite of the Otto Fiord Fm. of Lower Carboniferous age (in red in Figure 2), contained in diapiric bodies [5]. Ductile salt and evaporites rose and intruded the thick (12-15 km) sedimentary succession of the Sverdrup Basin [5]. The oldest exposed stratified rocks are Upper Permian to Middle Jurassic marine sandstones and shales, overlain by Jurassic to Cretaceous sandstones and limestones. The Mesozoic succession contains extensive volcanic units of basalt and sets of intrusive sills and dikes of diabase (intrusive basalt) of middle Cretaceous age (40Ar-39Ar dates from 129 to 80 Ma; [28]). The evaporite diapirs also contain rafts of diabase (Figure 2). Further afield are Cenozoic sediments of Paleocene to Eocene age, some disrupted by the mechanically intrusive evaporite diapirs, which may be actively rising [29]. Quaternary deposits consist of glacial and fluvial gravels and sands.

2.1. Eurekan Orogeny

The Paleozoic to Paleocene strata of the Sverdrup Basin were folded during the Paleogene Eurekan Orogeny into irregular synclines and anticlines with wavelengths of ca. 5 to 10 km. Distorted ramparts of diapiric anhydrite crop out in the cores of tight anticlines, and wider synclinal basins separate the diapiric walls. The most prominent faults are N-S, east-verging thrust faults that are interpreted by Jackson and Harrison [6] to be rooted in an evaporite canopy (extruded subhorizontal salt-anhydrite sheet) at depth. Important is a thrust fault south of Thompson Glacier at the head of Expedition River (rightmost side of Figure 2), which places Upper Permian rocks (Blind Fiord Fm.) and Carboniferous evaporite over Jurassic and Cretaceous strata. In addition, there are intermediate faults north of Expedition Diapir and Colour Peak Diapir, and the WGVA site is located at the intersection of a major WNW fault and a diabase intrusion of Cretaceous age (Figure 2). Minor faults and shattered zones (not indicated in Figure 2) abound around the evaporite diapirs, especially where ductility contrasts exist, such as between mafic igneous rocks and sedimentary rocks.

2.2. Wall and Basin Structure

Evaporite diapirs are conspicuous features in the Expedition Fiord area [25, 26]; they are clustered and irregular in shape, some forming walls and slivers [27]. Three clusters are identified in the local area from west to east: Colour Peak Diapir, Expedition Diapir (Gypsum Hill – Little Matterhorn Diapir), and a sliver east of Thompson Glacier (Figure 2). Early workers called the complexly folded region intermixed with encroaching diapirs a “wall-and-basin structure” (WABS; [30]). Schwerdtner and Van Kranendonk [31] described this feature as “crooked walls of diapiric anhydrite contained in folded clastic rocks.” The diapirs occupy the cores of narrow polygonal to oval anticlines. Jackson and Harrison [6] proposed that this restricted WABS province resulted from the development of a coalescing layer or canopy of evaporite rocks during the early Cretaceous; these evaporitic rocks would have plastically spread laterally and formed at depth the roots of now-exposed secondary diapirs created during compression, thrusting, and folding in the Paleogene (mainly Eocene) Eurekan Orogeny.

2.3. Diapirs and Springs

At the surface, the diapirs are composed of anhydrite with a carapace of porous gypsum with blocks of anhydrite, but they are known to have halite at depth from exploration drilling [32] and the presence of exposed halite at Stolz Diapir, located on southeastern Axel Heiberg Island [31]. Despite their susceptibility to erosion and the fact that they disrupt Quaternary deposits, their positive topographic relief (Figure 2) suggests continuing growth, which is being evaluated with satellite interferometry [29]. The position of perennial spring activity adjacent to or within evaporite structures led Nassichuk and Davies [32] to suggest an essential relationship between them.

2.4. Glaciers

The highlands of Expedition Fiord host the Müller Ice Cap, the largest on Axel Heiberg Island, crowning the Princess Margaret Range (2210 m.a.s.l.), ca. 40 km north of the area shown in Figure 2. One of the large outlet glaciers descending from it is the Thompson Glacier (Figure 2), spanning ~34 km in length and ~3 km wide in the ablation area. It dams several lakes, including Astro Lake (upper right corner of Figure 2) and Phantom Lake (2 km east of Astro Lake, outside Figure 2), and meets the White Glacier terminus at 80 m.a.s.l. (cf. [33]). White Glacier is a valley glacier occupying 38.5 km2, extending from an elevation of 80 m at its present terminus to its source on the mountains to the west at ca. 1782 m.a.s.l.; ice thickness exceeds 400 m in the deepest basins, and the glacier has thinned by approximately 20 m in the vicinity of the WGVA outcrop since 1960 [34]. Glacial activity has resulted in the carving of steep walls into Triassic-Jurassic sedimentary rocks, which in combination with contemporary thinning has exposed the array of hydrothermal veins described here.

3. Procedures

3.1. Field Work

Foot traverses by M.Z. were undertaken (2006) in the search for any evidence of hydrothermal activity that might help explain the anomalous thermal history results of the area (Table 1). A NE-trending ridge 2 to 3 km NE of the McGill Arctic Research Station (MARS camp, Figure 2), composed of diabase, shows evidence of alteration and gossans derived from oxidation of iron sulfides; the hydrothermal development described here (Figure 3(b)) was found at the intersection of the rusty ridge with White Glacier, which also coincides with the trace of a prominent fault or shear zone (Figure 2). Samples representing the various styles of mineralization were collected (2006, 2009) and prepared as double-polished thin sections (Dalhousie University). Carbonate deposits and fine-grained sediments associated with perennial spring discharge at Colour Peak Diapir (CPS) were collected by C.O. (2011, 2012). Carbonates were prepared as double-polished thin sections (Department of Earth Sciences, Western University); sediments were stored in native spring waters in air-tight serum vials capped with rubber stoppers until analysis.

3.2. Electron Microprobe

Wavelength-dispersive spectroscopic (WDS) electron microprobe analysis (EMP) on WGVA thin sections was completed at the R. MacKay Microprobe Laboratory (Department of Earth Sciences, Dalhousie University) using a JEOL 8200 microprobe operated at an accelerating voltage of 15 kV and a beam current of 20 nA and using a focussed beam (10 s on peak and 10 s off peak counting times; spot ). Qualitative energy-dispersive spectroscopy (EDS) was used initially for confirmation of optical microscopic phase identification.

3.3. X-Ray Fluorescence Mapping

Thin sections from both WGVA and CPS were analyzed by C.O. for trace elements by X-ray fluorescence mapping (μ-XFM) using synchrotron radiation (Advanced Photon Source, Argonne National Laboratory). Incident X-rays passed through a Si 111 monochromator on an insertion device (ID) at 13.5 keV to excite elements up to Pb. The incident beam was focussed to 5 μm × 5 μm using a Kirkpatrick-Baez mirror assembly. Spatially resolved XFM analyses were conducted over specific regions of interest in the samples, which were mounted 45° to the incident beam. The fluorescence X-rays were measured using a 12-element Canberra Ge (Li) detector. The fluorescence X-ray intensity was normalized by the intensity of the incident X-ray beam, and spectral analysis of the fluorescence spectrum of each pixel used to produce two-dimensional maps for each element analyzed.

3.4. Scanning Electron Microscopy

Thin sections and sediments from CPS were analyzed by C.O. by scanning electron microscopy (SEM) at the Western Nanofabrication Facility (Western University) using a LEO (Zeiss) 1540XB FIB/SEM with secondary electron (SE), backscattered electron (BSE), and in-lens detectors. Thin sections were coated with ~5 nm osmium metal, whereas sediments were initially washed in ddH2O and air-dried before coating with osmium. Samples were imaged at 1-2 keV at a working distance of 4-6 mm. Areas targeted for focussed ion beam (FIB) milling were coarse-milled at 10 nA followed by polishing at 1 nA using Ge at 30 keV. Sediments were also examined by transmission electron microscopy (TEM) at the Biotron (Western University) using a Philips EM 300 TEM. Samples were imaged as wet mounts on carbon-coated Cu grids at 80 keV.

3.5. Confocal Raman Spectroscopy

Confocal laser Raman microspectroscopy (LRM) was used to determine the composition of carbonate minerals, specifically to differentiate between aragonite and calcite polymorphs. LRM analysis was performed by J.H. using a Horiba Jobin-Yvon LabRam HR system at Saint Mary’s University. The instrument is equipped with a 100 mW 532 nm Nd-YAG diode laser (Toptica Photonics) and a Synapse charge-coupled device (CCD; Horiba Jobin-Yvon) detector. Pure silicon was used as a frequency calibration standard. Spectra were collected using an 80 μm confocal hole diameter, and a 600-groove/mm grating (spectral resolution of approximately ±2 cm-1). Spectra were collected using an accumulation of three, 50-60-second acquisitions with a laser spot size of ~1-2 micron at 100% laser power (~2.15 mW at sample surface). Spectra from different carbonate types were compared to those in Raman spectral libraries [35].

3.6. Fluid Inclusion Microthermometry

Fluid inclusion microthermometry and imaging was completed by J.H. and D.L. using a Linkham FTIR600 heating-freezing stage mounted on an Olympus BX51 microscope (Department of Geology, Saint Mary’s University). Stage calibration was carried out using synthetic fluid inclusion standards containing pure CO2 (melting at -56.6°C) and pure, critical-density H2O (melting at 0°C and homogenizing at 374.1°C). Uncertainties for these measurements are ±0.2°C when a heating rate of 1°C/min is used. Final ice melting temperatures were used to calculate salinities [36] and isochoric data were determined using the SOWAT software package [37, 38].

3.7. Stable Isotopes

Pyrite, quartz, and carbonates from vein margins and infillings of the WGVA were sampled using a ~0.5 mm tungsten carbide drill bit to produce powders. These powders were analyzed at Queen’s University Facility for Isotope Research (QFIR; Kingston, Ontario). Carbonates were reacted on a gas bench coupled to a Thermo Finnigan Delta Plus XP Isotope Ratio Mass Spectrometer (also using continuous flow). Oxygen from quartz was extracted using a conventional bromine pentafluoride extraction line, with the isotopes measured on a Finnigan Mat 252 Isotope Ratio Mass Spectrometer. Analytical uncertainties on the analyses are ±0.3‰ for O in silicates and ±0.2‰ for C and O in carbonates (K. Klassen, personal communication). Values are reported in the δ notation in units of per mil (‰) relative to Pee Dee Belemnite (VPDB) and Vienna Standard Mean Ocean Water (VSMOW), respectively. Stable sulfur isotope measurements were obtained using a Carlo Erba NCS 2500 elemental analyzer coupled to a Finnigan MAT 252 mass spectrometer with a Finnigan MAT Conflo 11. One sample of pyrite from the Colour Peak perennial springs was also analyzed. Values are reported in the δ notation in units of per mil (‰) relative to standard Vienna Canyon Diablo Troilite (VCDT). Replicate δ34SVCDT analyses are reproducible to within ±0.3‰ (K. Klassen, personal communication).

3.8. Trace Elements

Trace elements in carbonates were determined by J.H. using laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) at the Swiss Federal Institute of Technology (ETH Zurich). Aerosols were generated using a GEOLAS (now Coherent Inc.) 193 nm ArF excimer laser [39]. Aerosols were generated using a pulsed (10 Hz) laser beam with an energy-homogenized beam profile at a fluence of 15 J·cm−2 (80-90 mJ output energy) [40]. An Ar-He gas mixture (He 1.15 L/min; Ar 0.8 L/min) carried sample aerosols into an ELAN 6100 quadrupole ICP-MS using similar conditions as Pettke et al. [41]. Oxide production rates were maintained below 0.3%. Mass spectrometer dwell time was set to 10 ms for all masses measured. Quantification of trace element concentrations was performed using the software SILLS [42]. The standard reference glass 610 from NIST was used to calibrate analyte sensitivities. Stoichiometric Ca (40 wt %) was used as the internal standard value as carbonates were determined to be ~CaCO3 with minor impurities (<0.4 wt % total Fe + Mg + Mn).

3.9. Fission Track Thermochronology

The fission-track low-temperature thermochronological method as applied in this study at Dalhousie University has been fully explained in Grist and Zentilli [43] and references therein. It is based on the measured density and length distribution of linear tracks of crystal damage produced during spontaneous fission of trace amounts of 238U. In monotonically cooling systems, track densities provide a measure of age with respect to the closure temperature, which for apatite (a U-bearing calcium phosphate) is ca. 100°C over geologically meaningful time periods on the order of 10-100 Ma. Assuming a geothermal gradient of 30°C/km, this corresponds to a burial depth of 4-5 km. It is this temperature-dependent behaviour between 60°C and ~130°C (the partial-annealing zone or PAZ) that allows the apatite FT method to be used as a low-temperature thermochronometer. Reviews of the method, its application to geological problems, and the development of FT thermal models have been provided by Wagner and Van den Haute [44], Ravenhurst and Donelick [45], Gallagher et al. [46], and Gleadow and Brown [47]. All models were done using the Laslett et al. [48] annealing model; for the modelling, the accepted initial reduction in mean length to 15 μm was introduced (i.e., [49]). Data are shown in Figure 4 and Table 1.

4. Review of the Perennial Springs (PSS)

Examples of perennial groundwater discharge in the Canadian high Arctic have been documented at eight locations on Axel Heiberg Island [50, 51] and two on Ellesmere Island [52, 53]. While those on Ellesmere Island are linked to subglacial hydrology and geologic faults, saline perennial discharge having received the most attention occurs marginal to or near large evaporite diapirs [27] on Axel Heiberg Island, including those in Expedition Fiord (Gypsum Hill Springs, GHS; 79°2416N; 90°4352W; 98 m.a.s.l.) and Colour Peak Diapir (CPS; 79°2251N; 91°1609W; 3 m.a.s.l.) (Figures 2 and 3(a)) and nearby Strand Fiord Wolf Diapir (WS; 79°436N; 90°1239W; 132 m.a.sl.), not shown in Figure 2. Waters discharge either through numerous outlets or a single point source at constant rates and temperatures [3, 51], despite a mean annual air temperature of -15°C [1] and a lowest measured temperature of –54.6°C at a nearby weather station (Eureka, Nunavut, 1979) and measured permafrost thickness of 400–600 m [54].

Discharging spring waters are hypersaline, ranging from ~1 to 4 molal NaCl (4.7-18.9 wt% NaCl) [55, 56] (Table 2). Minerals including sulfates, carbonates, and halides have been documented in association with spring discharge, which form by a variety of disequilibrium processes including evapoconcentration and CO2 degassing [5558]. Icings (sheet-like masses of layered ice) form in distal areas during winter months when air and water temperatures are lowest, as well as carbonate and sulfate minerals thought to precipitate by freezing fractionation [51, 59]. At several sites, waters and sediments near spring outlets host microbial communities that are supported by carbon- and energy-rich reduced substrates including sulfur and methane [6065]. The source and activity of spring discharge at Expedition Fiord (CPS, GHS) has been generally interpreted to represent recent phenomena related to periglacial processes [7], but this model cannot account for activity at Wolf Spring, which is located south of Strand Fiord south of the area shown in Figure 2 [57].

Andersen et al. [7, 50, 66] proposed a qualitative and quantitative model for the PSS that envisioned a relatively surficial circulation of fluids, with recharge from Astro and Phantom Lakes, which are located on the eastern flank of Thompson Glacier, ~397 m above the outlet of the springs (upper right corner, Figure 2; Phantom Lake is located east of Astro Lake, just outside the map in Figure 2). The above authors assumed that the deep subsurface salt layer acts as a reservoir and conduit of water. Seasonally, the lakes drain partially (1-2%) and the water would circulate to the PSS through faults, such as the ones mapped in the area (Figures 2 and 3). Water would flow below the surface at a depth of 600-700 m via an evaporite layer, returning to the surface through the piercement structures associated with the springs. Below the surface, the water would attain the geothermal temperature (they assumed a geothermal gradient of 37.3°C/km after [54]). As it flows upward, the brines would lose heat to the surrounding permafrost.

This study brings a new perspective to the interpretation and significance of the PSS.

5. Results

5.1. Perennial Springs at Colour Peak Diapir (CPS)

Confirmed perennial spring activity is observed at three locations on western Axel Heiberg Island, two of which (Colour Peak and Gypsum Hill) are shown in Figure 2. They are among the most poleward perennial springs on Earth and are the only known example of cold, nonvolcanic springs in a region dominated by continuous permafrost. The two spring sites proximal to the hydrothermal development in Expedition Fiord are located adjacent to Colour Peak Diapir (CPS) and Expedition Diapir (GHS), with the springs at Wolf Diapir (WS) occurring ~60 km to the south.

Table 2 summarizes and compares the main geochemical and hydrological characteristics of these perennial springs. Groundwater emerges from either a single outlet or numerous seeps and springs (~40 at GHS, ~30 at CPS; [57]) resulting in total site discharge flow rates of 4–25 L/s. Discharge temperatures range from~ -6°C to ~ +6°C, but a review of past work suggests that outlet temperatures have remained constant (±2°C). The outlet waters have varying pH (5.8–7.4) but with high reducing potential (-171 to -312 mV) and low dissolved oxygen (0.2–0.5%). While concentrations of ionic constituents vary between sites, the waters are all Na-Cl-type brines (8-24% NaCl) with minor concentrations of Ca2+, Mg2+, K+, SO42−, NO3, and HCO3. Low concentrations of Fe2+, NH4+, and HS have been measured in the discharging waters.

A notable aspect that differentiates the CPS from GHS is elevated PCO2 in waters at CPS, which results in the oversaturation and subsequent precipitation of calcite (travertine) by CO2 degassing [55]. Travertine precipitation (Figure 5(a)) in narrow valleys form channels, which were initially described as being composed of alternating light (calcite spar) and dark (anhedral microcrystalline calcite combined with organic matter and noncarbonate minerals) laminae enriched in Fe, S, Na, and Si (Figure 5(b), [55]). Synchrotron radiation-based μ-XFM mapping across several of these micritic laminae (Figure 6) confirm the presence of Fe but also detect smaller concentrations of Mn, Zn, Cu, and Ni. A closer inspection of laminae within the calcite fabric by SEM shows two kinds of micritic fabrics: microcrystalline FeS grains and abrupt boundary layers (Figures 5(c)5(f)). FIB-SEM imaging of a region encompassing both micritic fabrics reveal trapped gas bubbles associated with cavitating brines, FeS grains embedded within the calcite matrix, and the boundary layer (Figures 5(g)5(j)). These microcrystalline FeS grains are similar in morphology to fine-grained sediment collected at spring outlets that are composed of small (0.1 μM) aggregates of FeS microcrystals (Figures 5(k) and 5(l)).

While it was previously suggested that Fe-rich laminations result from precipitation of iron oxyhydroxides [e.g., Fe (OH)3·nH2O; [55]], the combination of (1) accumulation of metals including Zn, Cu, and Ni; (2) morphological similarities between sediments observed in micritic laminate and spring outlets; and (3) the apparent absence of bacteria and/or biofilms in association with these sediments suggests that they are allochthonous. It is hypothesized that these sediments form as a by-product of subsurface sulfate-reducing bacteria that are subsequently entrained by groundwater and carried to the surface, where they accumulate in shallow outlet pools and are washed downstream and incorporated into the calcite fabric as micritic laminae during pulses of higher groundwater discharge.

5.2. Geological and Mineralogical Characteristics of the White Glacier vein array (WGVA)

The WGVA forms a conspicuous brown gossan exposed on a steep western flank confining White Glacier (79°26.66N; 90°42.20W; 350 m.a.s.l.), covering an area of more than 350 × 50 m and a range of elevation of ca. 50 m (Figures 3(b) and 7(a)). The veins are likely to be present as well under the glacier. The WGVA includes veins, mineralized breccias, and carbonate masses with a variety of structures (Figure 7). The host rocks are a dark grey chaotic mass of tight, matrix-supported angular breccia with angular fragments of sandstone, siltstone, shale, argillaceous limestone, dolomitic limestone, anhydrite, and gypsum of sizes predominantly <10 cm; the rock is reminiscent of fault breccia. According to the geological map, these beds belong to the Lower Jurassic “Savik beds” and the Upper Jurassic Awingak Formation [27], but the rocks are a jumbled mass, like clastic deposits near advancing salt allochthons described as off-diapir debris flows [6, 67]. These breccias match the description of anomalous, commonly overpressured rubble (termed “gumbo” in the Gulf of Mexico), found beneath allochthonous salt sheets and canopies in petroleum basins (e.g., [68, 69]).

Across from the mineralized zone, there is a wide band of highly sheared and broken-up rusty diabase (a conspicuous gossan), along an extensive Cretaceous diabase sill that can be followed for ca. 4 km along a steep ridge as far as Colour Lake, which lies adjacent to the MARS camp, near Gypsum Hill (Figure 2).

Mineralization consists predominantly of calcite, but textural and crystal habit evidence indicates that the earliest generation of Ca carbonate (cal1) originally crystallized as the polymorph aragonite (c.f. [70]). However, the aragonite composition is not preserved and has reverted to calcite on the basis of Raman spectroscopic measurements that show none of the characteristic vibrations for aragonite (142 cm-1, 162 cm-1, 190 cm-1, 211 cm-1, and 701 cm-1) and, despite crystal habit, the three dominant vibration lines for calcite are present (156 cm-1, 281 cm-1, and 712 cm-1) in cal1. Minor quartz, locally barite and abundant iron sulfides pyrite and marcasite (orthorhombic dimorph of pyrite), with minor pyrrhotite, chalcopyrite, sphalerite, and galena are present. All the structures suggest “open-space filling” with minimal replacement, reminiscent of epithermal and epigenetic deposits formed at shallow depths. Epidote and chlorite are common where the veins intersect diabase from the mafic dike (Figure 7(c)), and clay minerals are visible in thin sections. The early generation of brown calcite (cal1; after aragonite) occurs in crustified veins, generally as acicular crystals lining fractures, or in radial crystal aggregates filling cavities or around rock fragments (Figure 7(d)). A second generation of calcite (white-orange, sparry; cal2) forms massive infillings with minor quartz, in the central parts of some veins (Figures 7(b) and 7(d)). Locally, there are clear crystals of Iceland spar of up to 10 cm in size. A common structure of calcite is banded “zebra rock” consisting of acicular calcite layers alternating with darker, cm-thick black, carbon-rich sedimentary laminae (Figures 7(e)7(g)). The laminae are generally subhorizontal but locally contorted, showing variable inclined or vertical orientations in single exposures (Figure 7(f)), probably tectonically disrupted. Sulfides occur in darker laminae within the aggregates, in veins and as botryoidal aggregates filling large open cavities (Figures 7(g) and 7(h)). Sulfides visible in hand sample are pyrite and marcasite.

Under the reflected light microscope, pyrite occurs as subhedral to euhedral crystals of highly variable grain size (<1 mm to >1 cm), locally with cataclastic structure (Figures 8(a) and 8(b)). Marcasite was distinguished from pyrite by its conspicuous anisotropy and its commonly rosette or tabular habit, but in many samples both minerals occur together in rosettes, or plumose structures (Figures 8(c) and 8(d)) with alternating marcasite- and pyrite-rich bands. Some marcasite shows only faint anisotropy. Anhedral pyrite shows considerable porosity and minor remnant anisotropy, which may be the result of conversion from marcasite [71]; it probably crystallized as marcasite and was converted to pyrite in response to hydrothermal heating. Rare grains of galena, sphalerite (Figure 8(e)) and chalcopyrite, pyrrhotite, and arsenopyrite are observed as minute crystals hosted in early brown carbonate (cal1), locally close to the margins of the veins, within them or in the adjacent host. Pyrite framboids (5-8 μm in diameter and made of cubic or octahedral pyrite microlites of 1-2 μm) occur within black (organic-carbon rich) bands within laminated carbonate (Figure 8(f)). Some framboid aggregates appear to have partially recrystallized into euhedral pyrite (Figure 8(g)) but retain internal porosity filled with impurities.

Under the electron microprobe, pyrite (30 spot analyses) and marcasite (11 spot analyses) are compositionally indistinguishable and show a range of impurities, from below detection to maximum (in wt %) as follows: Mn (avg. 0.11%; max. 1.22%), Ni (avg. 0.03%; max. 0.23%), Co (<0.04%), Pb (avg. 0.06%; max. 0.26%), Cd (avg. 0.04%; max. 0.29%), Zn (avg. 0.01%; max. 0.08%), Cu (avg. 0.01%; max. 0.08%), Ag (avg. 0.002%; max. 0.02%), and As (avg. 0.003%; max 0.053%). Traces of Au were detected in some sulfide analyses but remain unconfirmed. Visibly zoned crystals () were analyzed at their centres and margins; while no consistent compositional patterns were detected, on average the centres have one order of magnitude more As, Mn, and Ni and are depleted in Cd. Apart from carbonates, qualitative evidence for the presence of titanite, barite, and other sulphates was found, as well as Fe and Ti oxides with varying amounts of Mn.

Pyrite framboids are difficult to analyze because the microprobe beam is larger than the microlites, and low-density interstitial material (possibly organic matter or volatile minerals) contaminates the analyses, and totals are low; thus, the results are only qualitative. Prominent impurities in framboid analyses () are Mn (avg. 0.15%; max. 0.60%), Pb (avg. 0.123%; max. 0.158%), Ni (avg. 0.09%; max. 0.19%), Cd (avg. 0.014%; max. 0.044%), Zn (avg. 0.006%; max. 0.021%), Cu (avg. 0.009%; max. 0.034%), and As (avg. 0.0175%; max. 0.053%).

One crystal of pyrrhotite was encountered with 0.06% Cu. One arsenopyrite grain was encountered that showed detectable Pb, Co, Ni, Zn, and Ag.

5.2.1. Trace Elements in Carbonate

Selected trace elements in cal1 and cal2 are listed in Table 3. Early brown acicular calcite has a hexagonal crystal form and elevated Sr, Ba, and Mg compared to sparry white-orange calcite. If abundant in the associated fluid, these ions are known to stabilize aragonite [72], which may have reverted to calcite as the hydrothermal fluid became diluted with time or warmer (c.f. [73]). Strontium and Ba are variable in concentration but reach concentrations as high as ~1000 ppm and ~770 ppm, respectively, with Sr and Ba showing a positive correlation, consistent with their precipitation from a fluid derived from dissolution of marine carbonates. In the absence of secondary dolomitization, lower Sr values emphasize meteoric water dilution. LREE/HREE ratios (e.g., PAAS normalized La/Yb) are ≪1 for carbonate analyses with , also diagnostic of hydrothermal carbonate precipitated from seawater or a fluid that interacted with, or dissolved, marine carbonate (c.f., [70]). Elevated concentrations of Ba likely indicate that small inclusions of the mineral witherite (BaCO3) are present as such high concentrations would not be present as dissolved Ba in the Ca-carbonate structure. Elevated concentrations of Pb, Zn, and Cu are likely not dissolved concentrations but rather indicate the presence of very small particles of included, coprecipitated or secondary sulfide minerals (e.g., galena, sphalerite, and chalcopyrite). Petrographic observation of trace quantities of these sulfides within veins and host rock margins or along host rock-vein contacts are in support of carbonate-sulfide coprecipitation.

5.2.2. Fluid Inclusion Petrography and Microthermometry

Carbonates from representative vein specimens (discrete veins, and “zebra” texture) from the WGVA were examined for fluid inclusions. The vein samples selected for study (e.g., Figures 9(a)–9(c)) contain an early generation of zoned, brown- to tan-coloured, acicular, hexagonal early brown calcite (as cal1) that grew inward from fracture walls to create a symmetrical open-space filling layer on the walls, and a later generation of white to orange massive, polycrystalline to euhedral, terminated “sparry” white-orange calcite (cal2) that infills the central vein cavities lined by terminated crystals of cal1. Quartz is also present, but rare, occurring as a massive, polycrystalline infilling within an even later cavity intergrown with cal2. No differences in the petrographic characteristics (inclusion origins, morphologies) of fluid inclusions from the White Glacier and Striae Hill study areas are observed. The infilling cal2 generation contains primary and secondary inclusions. Both generations of calcite post-date an early generation of base metal sulfide mineralization, with zoned cal1 coating, and infilling fractures within pyrite-marcasite (Figure 9(d)). Trace amounts of secondary galena, pyrite, and sphalerite grew synchronous to, or after, cal2. Primary inclusions in cal2 (Figures 9(e)–9(i)) are rare, up to 80 μm in longest dimension, and have a variety of morphologies, ranging from complex, rectangular to rhombic (Figures 9(f)–9(h)) to long, rectangular (tube-like) in shape (Figure 9(i)). The primary inclusions (type “P”) occur in isolated assemblages of up to ~10 inclusions within clear domains of calcite. Secondary inclusions are smaller, and at least two different types are recognized. Type “S” inclusions (up to 30 μm; averaging ~15 μm) (Figures 9(e)–9h, 9(k), and 9(l)) show morphologies ranging from amoeboid, subangular, through rectangular to rhombic, some occurring in planar arrays that cross-cut calcite grain boundaries and cleavage planes. Type “S2” inclusions are less abundant, very small (<4 μm), and have consistently rectangular to rhombic morphologies (Figure 9(m)). The wall-lining cal1 generation contains only irregularly shaped secondary inclusions (Figure 9(n)). In both cal1 and cal2 generations, all inclusion types are two-phase at room T, containing an aqueous liquid () and a vapour bubble (V), with L : V ratios varying from ~85 : 15 to 90 : 10. Fluid inclusions in the late quartz are too small to investigate (<5 μm, and obscured by a lack of transparency in the quartz) but appear to be two-phase .

Table 4 and Figures 10(a)–10(c) summarize microthermometric and calculated salinity data for inclusions in representative secondary fluid inclusion assemblages (FIA) within cal1 and primary and secondary assemblages within cal2. A total of 120 fluid inclusions were measured. In some FIA, only partial microthermometric data could be obtained owing to difficulties in accurately observing ice melting or final homogenization. Primary fluid inclusions show final ice melting (; from two sites) between -7.2°C and -16.8°C, corresponding to bulk salinity between 10.7 and 20.1 wt% NaCl equivalent. Average bulk salinity for primary inclusions is 15.0 wt% NaCleq. Total homogenization () of primary inclusions occurred by vapour bubble disappearance (), between 153°C and 293°C (; ). Secondary type “S” fluid inclusions show a typically much wider range in final ice melting () between 0°C and -16.9°C, corresponding to bulk salinity from 0 to 19.8 wt% NaCl equivalent. However, high-salinity secondary inclusions are rare, and the average salinity for secondary inclusions is much lower (3.7 wt% NaCleq.; ). The range in for secondary inclusions is similar to that for primary inclusions, between 112°C and 308°C (; ). Overall, there are no discernable differences in salinity or Th data between secondary inclusions in cal1 or cal2. This, combined with the near complete lack of very high-salinity secondary inclusions in cal1 (as is typical of primary inclusions in cal2), suggests that secondary inclusions in cal1 and cal2 contain the same fluids. Figure 10(a) ( primary, 68 secondary) shows no correlation between salinity data but illustrates a continuum of inclusion salinities from primary inclusions with consistently highest salinities, through a wide range in secondary inclusion salinity approaching and including ~ pure water. There is relatively little intra-FIA variability in inclusion salinity with only a small number of FIA showing anomalously high- or low-salinity inclusions (Figure 10(b)); given the range in inter-FIA salinity observed in secondary inclusions, anomalously high- or low-salinity inclusions within single assemblages may reflect incorrect assignment for some inclusions within larger groups to a given FIA. Compared to the salinity data, data (Figure 10(c)) shows consistently large intra-FIA variability, up to a ~150°C variation in single FIA. It was noted that stretching of inclusions during heating experiments had a negligible effect on ; repeated heating experiments on single inclusions showed no more than a ~1°C increase in even after holding above for 1 hour before repeat measurements. Petrographic scrutiny of inclusions before microthermometry was performed to exclude any inclusions showing obvious evidence of necking down. However, in the absence of any detectable visual evidence for post-entrapment modification, the effect of inclusion leakage and necking down remains likely given the range in in some FIA. Variability in must therefore reflect post-entrapment leakage or necking down (that could not be detected optically) or real variations in trapping conditions (T or P).

Sparry white calcite contains high-salinity primary inclusion assemblages with individual inclusion salinities ranging from 10.8 to 20 wt% NaCl and values between 153°C and 263°C. Secondary inclusion groups show a wider variation in salinity from 15.6 wt% to as low as ~0 wt% NaCleq (within measurement uncertainty of ±0.2°C, equivalent to salinity uncertainty of <0.2 wt% NaCleq.) and values between 112°C and 251°C. Importantly, the transition to lower salinities from primary to secondary inclusions suggests overall dilution with time, though fluctuations in salinity during secondary inclusion entrapment appear large (i.e., Figure 10(c), secondary inclusions in cal1 and cal2).

5.2.3. Stable Isotopes

Oxygen and carbon isotope values in carbonate generation cal1 (Table 5, Figure 11) range from +1.4‰ to +3.5‰ (δ18OVSMOW) and -22.5‰ to -31.2‰ (δ13CVPDB), respectively. These data are similar to that for cal2 ( to +3.2‰, and to -23.8‰). Drusy quartz coeval with cal2 has . In comparison, the modern (active) spring carbonate has significantly higher isotope ratios ( and ). The isotope composition of the modern spring carbonate lies close to the compositional range for post-mineralization pore- and fracture-filling calcite in MVT deposits in the Canadian Arctic (Figure 11; [74, 75]). With respect to O isotopes, at a °C (mean annual spring discharge T), we calculate a value of (line “B” in Figure 11) for the aqueous solution in equilibrium with the carbonate, slightly higher than the composition of high-latitude meteoric water (i.e., melt water) for the study area (-26‰; OIPC3.1 calculator, University of Utah; line “B” in Figure 11). Over a temperature range from 150 to 300°C (based on primary fluid inclusion ) and using a meteoric water composition of (negligible change in study area latitude beginning in the Eocene-Miocene when the paleosprings at WGVA are assumed to have been active), calculated values of δ18OVSMOW the carbonate and quartz expected to precipitate from that water range from -22‰ to -14‰ and -19.5‰ to -11‰, respectively (fields “C” and “D” in Figure 11). Note that these ranges represent maximum values since values are minimum fluid inclusion (host mineral precipitation) minimum fluid inclusion entrapment temperatures. Importantly, these values are at least ~6‰ lower than the measured mineral compositions. Estimated δ18OVSMOW values for fluid in equilibrium with cal2 are lower than for cal1 (Table 5) assuming cal1 precipitated at a similar or higher than cal2. Consistent with the fluid inclusion salinity data, these results suggest that the hydrothermal minerals did not precipitate from heated meteoric water but rather from a mixture of meteoric water and a saline fluid phase, with a larger proportion of meteoric water in the mixture at the time of precipitation of the late carbonate-quartz infilling. Thus, while modern springs show a predominance of meteoric water, the paleosprings formed from mixtures of meteoric water and basinal-type brine, with fluctuations from a dominant brine to dominant meteoric water influence with time. This transition is consistent with fluid inclusion data, showing decreasing salinity from primary to secondary inclusions, approaching ~pure water in many inclusion assemblages.

The strongly depleted nature of carbon isotopes for the WGVA vein carbonates (δ13CVPDB in cal1 ranges from -22.5‰ to -31.2‰, and that in cal2 ranges from -8.4‰ to -23.8‰) likely reflects a dissolved carbonate source related to oxidation or thermal degradation of organic matter. A shift to higher δ13CVPDB values from cal1 to cal2 is observed, consistent with the shift from higher to lower values of calculated δ18OVSMOW for the fluid, resulting from a combination of progressive dilution by meteoric water and/or decreasing equilibration .

Values of δ34SVCDT in pyrite(-marcasite) in the vein margins of the WGVA are variable (-2.7 to +16.4‰) with the highest value approaching that for Carboniferous-age evaporite diapir sulfate [76]. The wide range in δ34SVCDT, combined with textural evidence for recrystallization of pyrite framboids, and likely thermal and compositional fluctuations in the hydrothermal system with time suggest that S isotope systematics of the WGVA reflect a combination of preserved low-temperature microbial sulfate reduction and higher-temperature reduction of sulfate by organic compounds (c.f. in MVT systems; [77]).

6. Discussion

6.1. Justification for an Alternative Model

The surficial model by Andersen et al. [7], which envisaged recharge from glacial lakes and near-surface circulation through evaporites and permafrost (see Section 4), was justifiable with the data available, but the younger thermal history detected by thermochronology suggested that the PSS phenomenon might have a deeper, internal energy connection. The apatite fission track dating method (Figure 4) constrains not the age, but the geological time when rocks cooled for the last time below ca. 100°C (corresponding to a depth of 3 to 4 km under normal geothermal gradients) and allows for modelling compatible time-temperature histories (e.g., [78]). Regionally, rocks now exposed at the surface cooled to temperatures below 100°C during uplift and exhumation due to the rise of mountains on Axel Heiberg Island and Ellesmere Island (Figures 4(a) and 4(b)) between ca. 62 and 33 Ma (Paleogene; [79]) caused by tectonic collision with Greenland (e.g., [5, 80]). In contrast, rocks in the Expedition Fiord area of Axel Heiberg Island (Figure 4(c)) cooled 10 to 20 My later [9, 81].

Two samples are clearly anomalous (Figure 4(a)): (1) sample SF-66, collected at the surface of the Cretaceous age gabbro Wolf Intrusion (Figure 2) between Colour Peak Diapir and White Glacier, and (2) sample SF-59, collected in a raft of gabbro intrusive within the Colour Peak Diapir (Figure 2). Their apatite fission track ages are younger than most regional samples, and their mean track length is the shortest measured in surface rocks in this area (Figure 4(a); Table 1). These characteristics suggest that, relative to other rocks, they remained within the crust for a longer time at a higher temperature. Figures 4(b) and 4(c) depict the result of Monte Carlo inverse thermal history models for two representative samples, including that from the Wolf Intrusion (SF-66) and another (SF-28) from near the summit of the Princess Margaret Range, ca. 60 km to the northeast (79.7°N; 88.5°W; 2467 m.a.s.l.). Both internally consistent cooling models used the AFTINV annealing model [48, 82, 83]. The model in Figure 4(c) is compatible with the interpretation that the rocks now at the surface in the region of the Wolf Intrusion (sample SF-66, Figure 2) could have maintained temperatures of up to ca. 75°C until the Miocene (ca. 15 Ma) and cooled since. Acceptable model solutions involving heating at higher temperatures for shorter time periods may also be compatible with the data (Figure 4(c)).

6.2. Comparison of PSS and WGVA

The active PSS in Expedition Fiord and the high-temperature veins in WGVA have many geological similarities that suggest a kinship, calling for a modification of the previous interpretations that the PSS are merely recent periglacial phenomena (e.g., [7, 50]).

6.2.1. Location

The PSS are in proximity (marginal) to anhydrite-gypsum diapiric structures, in zones that are characterized by faulting and brecciation. They are close to sea level or at the level of local rivers. The WGVA are not so close (~1-3 km) to mapped evaporite diapirs (Figure 2), but the brecciated evaporitic host rocks exposed and the proximity to large faults, which are probably rooted in diapirs at depth as suggested by geological cross sections by Jackson and Harrison [84], make it likely that a portion of their postulated evaporite salt canopy structure exists in the subsurface.

6.2.2. Size

The exposed plumbing systems (Figure 3) are comparable in areal size: PSS at Colour Peak Diapir and Gypsum Hill Diapir are 250 × 200 m and 400 × 50 m, respectively, but it is probable that they extend underneath Expedition Fiord and the Expedition River valley. The WGVA has exposures of more than 300 × 50 m but also extends under the glacier. Its visible vertical extent is 100 m or more; an unknown volume has been eroded.

6.2.3. Structures and Textures

The plumbing system at the WGVA consists of predominantly vertical, vein-like conduits lined by radial aggregates; however, both the WGVA and the PSS at Colour Peak Diapir exhibit banding, cyclic laminae at the mm scale in the PSS and mm to cm scale in the WGVA. The WGVA veins are crustified, often symmetrical, with acicular or fibrous crystals growing perpendicular to the margins, and some have large sparry (clear) crystals in the centre. The cyclicity in both implies the action of recurring pulses of mineral-laden fluids, which is the likely cause for laminated fabrics in the PSS mineral precipitates at Colour Peak Diapir. Sulfides occur in the WGVA within microfractures in the host rocks and as botryoidal aggregates in former open spaces up to decimetre scale.

6.2.4. Mineralogy

Both systems are mineralogically similar. The predominant mineral in both the PSS at Colour Peak Diapir and the WGVA is calcium carbonate in the form of calcite; laminated fabrics at Colour Peak Diapir contain mixtures of carbonate and micritic laminae with organic matter [55] and metal enrichments in Fe, Mn, Zn, Cu, and Ni. In addition to a predominance of calcite (CaCO3), other phases identified in the PSS are halite (NaCl), ikaite (CaCO3·6H2O), gypsum (CaSO4·2H2O), thenardite (Na2SO4), mirabilite (Na2SO4·10H2O), and elemental sulfur (S°), which form in association with spring discharge [5557]. The calcite in the WGVA shows signs of recrystallization judging from its mineral habit (orthorhombic blades and pseudo-hexagonal prisms resulting from twinning); therefore, it probably crystallized originally as aragonite. The pseudomorphs, however, yield calcite Raman spectra. Sulphates are also present as anhydrite (CaSO4), gypsum, and barite (BaSO4) but in the host breccias, not in the veins. The Fe- and Mg-rich mafic igneous host rocks of the WGVA contain epidote and chlorite as alteration of magmatic silicates, and the bulk of the Fe sulfides (marcasite and pyrite) occur within or near the mafic rocks, suggesting that they supplied the iron. However, the “zebra” banded carbonates rich in organic matter contain sulfides in the form of framboids and botryoidal aggregates. Sulfide (pyrite) microcrystalline aggregates similar to framboids are also observed within the banded mineral precipitates of the PSS at Colour Peak Diapir (Figure 5); the presence of hydrogen sulfide in discharging waters suggests microbial sulfate reduction, which is confirmed by sulfur isotope data of micritic sediments (see below). Chalcopyrite, sphalerite, galena, pyrrhotite, and arsenopyrite have so far been observed only in the WGVA and especially in the proximity of mafic igneous rocks.

6.2.5. Fluid Chemistry

A direct comparison of the chemistry of the rocks in the PSS and WGVA is not possible because the sampling and methodologies utilized were different. However, the fluids within the former can be compared with the inclusion fluids in the latter. The salinity of the brines in the PSS is bimodal, some yielding relatively low salinity fluids between 8 and 9% and others higher-salinity fluids of 16-17%; similarly, there is evidence for two different types of fluids in the fluid inclusions in the WGVA, 1.5% and 16% NaCl equivalent. The inclusion data in the WGVA suggest a progressive dilution with time from a ~20 wt% NaCl brine end-member down to ~pure (meteoric) water with evidence of large variability in salinity during progressive trapping of secondary inclusions. In comparison, the modern springs in the PSS show a variability in salinity from location to location from ~4.7 to 18.9 wt% NaCl equivalent [57] supporting the kinship between both systems. In both systems, chlorine is the most abundant halogen, and Ca and Na are the most abundant alkalis, reflecting the spatial association with evaporites, also manifest by the presence of sulfate. More internally consistent work needs to be done to confirm the comparisons made here.

6.2.6. Metals

Iron is the predominant metal sulfide in both systems, with 35 to 100 ppm sulfide (presumably as Fe sulfide) in the PSS precipitates [57], although the surfacing brines are not enriched in Fe (2 ppm; [60, 63]). In the WGVA, the sulfide content is extremely variable, from traces to 100%. The geochemical analyses of heavy mineral concentrates contain 32% Fe, 0.2% Mn, and small amounts of Ag (0.2 ppm), Cu (9 ppm), Pb (3 ppm), Zn (13 ppm), and Ba (430 ppm). These values are like those detected in 3 rock samples from a gossan in a raft of diabase adjacent to the PSS Colour Peak Diapir.

6.2.7. Stable Isotopes

Oxygen isotopes: the δ18OVSMOW value of the PSS hypersaline fluids is +10‰, thus dominantly high-latitude meteoric water but with a very minor non-meteoric (brine) component. Recalculation of the water composition that would be in equilibrium with the perennial spring carbonate at ~6°C (Figure 11, Table 5) gives a δ18OVSMOW value of -22.1‰ (similar to -26‰ for modern meteoric water in the study area; constrained using lat-long-elevation data using the OIPC3.1 calculator, University of Utah) suggesting a greater meteoric water contribution to the existing spring carbonates than in the current circulating fluid. The δ18OVSMOW values in the early brown (cal1) and later (cal2) carbonates at the WGVA range from -0.3‰ to +3.5‰, which based on fluid inclusion minimum trapping temperatures (~150-300°C), constrain minimum δ18OVSMOW values for coeval fluid between -0.2‰ and -8.7‰. This range in fluid δ18OVSMOW is explained through mixing of high-latitude meteoric water and a heated saline formational water or basinal brine.

Sulfur isotopes: (δ34SVCDT) in the PSS at Colour Peak Diapir is +19.2‰ (sulfide separated from laminated carbonate-organic matter deposits) consistent with microbial sulfate reduction. In the WGVA, they range from -2.7‰ to +16.4‰, having a more ambiguous origin but likely preserving either low-temperature microbial sulfate reduction or higher-temperature reduction of sulfate by organics. The sulfur isotope systematics may have been somewhat altered by surface weathering, and more research is needed to confirm that the range in values for sulfides from the WGVA are not modified.

Carbon isotopes: δ13CVPDB in the PSS-laminated carbonate at Colour Peak Diapir is -10‰, suggesting an organic carbon component reflected in the measured isotope composition. The δ13CVPDB values in carbonates in the WGVA are -20.6 to -31.2‰, compatible with an origin from degradation of petroleum at depth. Specifically, the highly negative δ13CVPDB values associated with early cal1 are consistent with sourcing of hydrothermal carbonate from fluids that oxidized organic matter. Increasing δ13CVPDB with time approaches the same value as the PSS carbonate and nearer to the compositional field for post-ore carbonates filling pore spaces in host rocks from major MVT deposits in the Arctic of Canada [74, 75].

6.2.8. Temperature

As indicated above, the temperature of the PSS at Colour Peak Diapir is rather invariable at ~6°C [55]. The fluid inclusions in the WGVA indicate that the fluids had temperatures from 100°C to 300°C (Figures 10(a) and 10(b)). This higher temperature is compatible with the presence of epidote adjacent to the veins, requiring temperatures between ~200°C and 350°C to crystallize from hydrothermal fluids [85, 86]. It is unclear whether the fluids at depth in the PSS have warm temperatures and equilibrate with the surrounding permafrost to result in a temperature of ca. 6°C as suggested by Andersen et al. [7] or if they presently circulate to greater depths, as they must have done in the past to account for the fluid inclusion data. The inclusion data do not portray any correlation between and salinity across the large continuous range in salinity observed. This can be interpreted in 2 ways: (a) maintenance of fluid T (i.e., between approximately 150°C and 250°C) during progressive dilution (unlikely because it would require the low-salinity fluid to be heated before mixing) or (b) a cooling trend from early high salinity to low salinity, if P decreased from quasi-lithostatic to hydrostatic conditions (i.e., if we correct for P); exhumation of the system may have contributed to this transition. This second explanation agrees with stable isotope data (showing increasing amounts of likely meteoric water requiring a more open system to allow incursion and mixing).

6.3. Framboidal Pyrite

Framboidal pyrite has been observed in the WGVA (Figures 9(f) and 9(g)), and Fe-sulfide microlite-like constituents of framboids are observed in micritic layers in banded carbonate in the PSS at Colour Peak Diapir (Figure 5). The origin of framboidal pyrite has been debated since first described from ores in sedimentary strata by Schneiderhöhn [87] who described them as “mineralized bacteria,” although subsequent research demonstrated that they can form in purely inorganic systems. Nevertheless, experimental work shows they generally form at temperatures below 100°C in environments where bacteria are present. Wilson et al. [88] argued for bacterial growth of framboids within liquid petroleum at depths of several km, based on textural evidence and sulfur isotopes, and recently Lin et al. [89] demonstrated that framboids can grow rapidly where anaerobic oxidation of methane is coupled with microbial sulfate reduction. The spherical shape of framboid aggregates has been ascribed to colloidal phenomena, or to the replacement of minute vacuoles, but their formation has proven extremely difficult in the absence of organic matter [90]. The presence of organic matter enhances the chemical formation of pyrite combining metals and biologically produced hydrogen sulfide [90]. Wilkin and Barnes [91] explain the growth of framboids in four consecutive stages: (1) nucleation and growth of initial iron monosulfide microcrystals, (2) reaction of the microcrystals to greigite (Fe3S4) or mackinawite (Fe9S8), (3) aggregation of uniformly sized greigite microcrystals, i.e., framboid growth, and (4) replacement of greigite framboids by pyrite or marcasite. Some of the microcrystalline Fe-sulfide textures observed in the PSS at Colour Peak Diapir (Figure 5(k)) illustrate this sulfide evolution. Vietti et al. [92] found that framboids grew readily within sulfidic microbial biofilms by bacterial degradation of organic matter. If there is continued influx of Fe, the originally spheroidal aggregates of microlites progressively recrystallize into idiomorphic (cubic or octahedral) crystals; this recrystallization is prevented if the framboids are isolated from new input of Fe, as within a biofilm, petroleum, or an enclosing mineral [93, 94]. Pyrite framboids formed during catastrophic dysoxic events in sedimentary basins [95] and in methane seeps associated with gypsum, where biogenic activity is possible [89]. Despite controversy, some authors are persuaded that pyrite framboids are associated with bacterial activity (e.g., [96, 97]). Both low-grade metamorphism and fluid-rich diagenesis can lead to framboid recrystallization and formation of aggregates with framboidal cores and euhedral edges [98]. Heating above 150°C makes marcasite revert to pyrite [49, 99, 100].

The presence of pyrite with microlites similar to those in framboids in the PSS at Colour Peak Diapir is a possible sign of anoxic reduction during the flow of hydrocarbons (methane) and/or microbial activity, which would explain the preliminary sulfur isotope signature. The presence of framboidal pyrite in the WGVA is suggestive of interaction with hydrocarbons at depth but also of microbial activity when marcasite was also the stable phase of Fe sulfide, probably at temperatures below 100°C. The evidence for recrystallization of framboids and the reversion of marcasite to pyrite indicate subsequent heating.

6.4. “Zebra” Textures

The “zebra”-textured veins at WGVA (Figures 7(e)7(g)) were formed by “open space filling” and have similar mineralogy and texture to subvertical veins in the area, within diabase or brecciated sedimentary rocks. The term “zebra” texture has been applied to arrays of subparallel veins in base metal ore deposits or dolomitized carbonate masses, where there is light and dark banding that reminds of a zebra coat. They have bilateral symmetry about a central mineral band (C) with repetition of sulfide and/or gangue mineralogy in the following form: A B C B A, where A, B, and C represent different mineral phases or different generations of a single mineral phase (c.f. [101103]). Originally, sedimentological and diagenetic evidence led early authors to describe their formation immediately below the sediment-water interface as “diagenetic crystallization rhythmites.” However, the early-diagenetic interpretation has been rejected as a general model, since various types of zebra veins have been described in various geological environments. For instance, in the Nanisivik Pb-Zn-Ag deposit located on northwest Baffin Island (lower right in Figure 1), zebra vein arrays very similar to those at WGVA were interpreted to have formed by pulses of basinal fluids, later than lithification of the host sediments at temperatures more than 200°C [104, 105]. Recent workers (e.g., [106]) interpret zebra veins in a Mississippi Valley Type base metal deposit in the Andes as having formed as sheet cavity networks developed by extensional fracturing in deep subsurface environments such as the filling of subhorizontal tension fractures formed during thrusting. Under these conditions, the most effective mechanism to generate subparallel tension fractures is fluid overpressure and episodic hydraulic fracturing (e.g., [107109]), repeated cycles of hydraulic fracturing and healing like the mechanism responsible for banded veins referred to as “crack-seal” veins (e.g., [110]). The generation of episodic basin dewatering of overpressured fluids in sedimentary basins with evaporites is a common phenomenon, especially if enhanced by compressive tectonics [111113], but also due to rapid compaction [114] and in the environment of salt dome cap rocks [115]. Finally, Warren [69] discusses the common development of geopressured fluids in the environment of salt canopies similar to the one underlying Expedition Fiord as defined by Jackson and Harrison [6]. An alternative hypothesis for the formation of zebra textures relates them to force of crystallization during diagenesis or mineral replacement (e.g., [102, 116]), but we consider these interpretations not applicable to the WGVA.

Similar “zebra” textures in the PSS at Colour Peak Diapir (Figure 5(b)) at a millimeter scale were formed by a different process, since they are obviously formed at or near the surface by successive pulses of expelled fluid. It is likely that at depth in the PSS (and analogous WGVA) the conduits are lined with similarly banded, crustified mineral coatings. Drilling will be necessary to better understand their nature.

6.5. Towards a Deeply Circulating Model for the PSS

The similarities between the PSS and WGVA make it likely that both systems have some genetic kinship. Pollard et al. [3] considered it unlikely that meteoric or surface water could be the source of the brines in the PSS, yet the hypothetical model by Andersen et al. [7, 50, 66] (Figure 12(a)) envisaged relatively surficial circulation of fluids, with recharge from Astro and Phantom Lakes, which are located on the eastern flank of Thompson Glacier, ~397 m above the outlet of the springs (upper right corner of Figure 2). Andersen et al. [7] assumed (Figure 12(a)) that the deep subsurface salt layer acts as a “reservoir and conduit” of water. Seasonally, the lakes drain partially (1-2%) and the water would circulate to the PSS through faults, such as the ones mapped in the area (Figure 2). Water would flow below the surface at a depth of 600 to 700 m via an evaporite layer, returning to the surface through the piercement structures associated with the springs. Below the surface, the water would attain the geothermal temperature (they assumed a geothermal gradient of 37.3°C/km after [54]). As they flow upward, the brines would lose heat to the surrounding permafrost.

The Andersen et al. [7, 50, 66] model is problematic. Firstly, evaporites (salt, anhydrite, and gypsum) in the near-surface are among the most impermeable rocks in nature. Powers et al. [117] and Holt and Powers [118] have argued that it is virtually impossible for meteoric water to penetrate evaporitic rocks; under similar conditions in salt (halite), water would take 3 to 30 million years to advance 1 m, in anhydrite 300,000 years [119]. Even in active salt springs around salt diapirs in semi-desert Iran, Zarei and Raeisi [120] assume that surface recharge cannot penetrate more than 10 m. A further problem is that if halite is present, it is difficult to maintain permeability, because porosity becomes occluded [121]. Peripheral fracturing can lead to focussed fluid flow (e.g., [69]), but the geological map and cross sections (Figure 2; [6, 27]) show that there is no likely shallow continuous body of evaporite between Astro Lake and Colour Peak Diapir, a distance of ~20 km. More likely evaporites act as impervious barriers, and any circulation would be peripheral and focussed by them. The surficial water flow model is unlikely considering the independence of PSS brine flow temperature and wide air temperature variations between winter and summer at this latitude. Evaporite-focussed fluid leakage was postulated by Keen [122] to explain heat flow anomalies above salt diapirs in offshore Nova Scotia. Although salt conducts heat many times more effectively than other sediments, Keen [122] concluded that conduction alone could not explain the anomaly, and advective heat from escaping warm fluids was required; for the model, she assumed that the aquifer temperature would be 150°C, hence definitely did not consider surficial seawater. Evans et al. [123] have explained the mechanisms responsible for advection of fluids near salt domes, and Warren [69] discusses the flow of warm brines peripheral to evaporite diapir regions where salt canopies have developed. In the Expedition Fiord area, the evidence is compatible with models in which the PSS fluids migrate upward from considerable depth following previously developed plumbing systems marginal to evaporite bodies (Figure 12(c)), either by vein arrays such as those at WGVA or brecciation typical of the margins of displaced allochthonous salt masses [69]. The involvement of hydrocarbons in the reactions leading to precipitation of minerals including carbonates and sulfides in the above systems further suggests a connection with petroleum seeps, difficult to ascribe to surface recharge, especially in a polar region with thick permafrost. Liquid and gaseous hydrocarbons are present in fluid inclusions (Zentilli, unpublished data) in the Colour Peak Diapir (Figure 1), and deep wells drilled on salt-cored anticlines indicate the presence of adequate source rocks [21].

The apatite fission track data suggest that the rocks now at the surface in Axel Heiberg Island outside the area of the WGVA were at temperatures of ca. 100°C in the Eocene (Figure 4(b) and [8]). Based on industrial thermal data (156 deep wells) and coalification gradients, during the Paleogene the geothermal gradient in this region is estimated to have been 28 ± 9 mK/m [124], which translates into 2.7 to 5.3 km (avge. 3.6 km) of erosion. Since in the area of the PSS and WGVA the thermal regime was anomalous due to fluid advection, it is problematic to be certain of its paleo-geothermal gradient. However, the apatite FT models are compatible with rocks now at the surface having been at temperatures of ca. 75°C (with large uncertainty, Figure 4(c)) until the Miocene, and one can estimate between of 0.5-2 km of exhumation until the veins were exposed by glacial erosion. The WGVA is located near the bottom bedrock of the White Glacier valley; hence, much of the mineralized zone may have been eroded. It is problematic to assess how much pre-glacial erosion has taken place in the Expedition Fiord area. When mountains in Axel Heiberg Island were rising in the Eocene, the climate was warm and humid, leading to the formation of forests (e.g., [10]), and it is unlikely that large glaciers occupied the region through the Miocene, so ensuing erosion was probably due to fluvial incision and exhumation related to landsliding of steep slopes (John Gosse, personal communication). The Late Miocene through Pliocene appears to have been a time of net deposition in the Canadian Arctic, resulting in the infilling of pre-existing valleys or grabens, and extensive depositional surfaces until 2.7 Ma [125127]. Those surfaces were subsequently incised, perhaps initially by streams, but more significantly by warm-based outlet glaciers [128] and independent valley glacier systems. While erosion by valley and outlet glaciers through unconsolidated sediment is likely to have been rapid, even bedrock has been incised by more than 80 m/Myr [129] to form fiords. However, under cold-based glaciers which occupied many summit plateaus in Arctic Canada, subglacial erosion can be negligible, as indicated by the survival of moss under cold-based glaciers for more than 5 ka [130, 131].

The fluid inclusions indicate the carbonate and sulfide minerals in the veins crystallized from more than one pulse of hot fluids with temperatures between 100°C and 300°C, and strong swings in pressure can be ascertained. The veins and zebra textures were formed in open spaces, which at a depth of a few km could only have been present if the fluids were at lithostatic pressure plus the tensional strength of the rocks [107]. The environment in which the WGVA developed is one of evolving allochthonous evaporite, with a large salt canopy at depth [6]. The region underwent strong compression during the Paleogene Eurekan Orogeny (Figure 12(b)). The zebra textures described at the WGVA are like zebra textures developed in compressive tectonic environments associated with thrusting in fluid overpressured zones (e.g., [106]). The host rocks of the WGVA are fractured mafic igneous rocks physically intermixed with a breccia or conglomerate that has some similarities with “gumbo” found in association with displaced allochthonous salt masses in various basins rich in evaporites in Oman, the Gulf of Mexico, offshore Angola and Brazil; these brecciated rocks are characteristically strongly overpressured when intersected by drilling [69].

Although overpressures may develop under autochthonous and evolving allochthonous layers of salt by simple compaction [132, 133], in some instances it leads to the generation of metallic ore deposits [114]; compressional tectonics can intensify them (Oliver, 1986; [112]). The Eurekan Orogeny (Paleogene) in the Sverdrup Basin resulted from collision of Greenland with the Canadian Arctic Archipelago [134] and led to the formation of high mountains (Figure 1) by stacking. Most thrust faults are rooted in salt, and a particular case is the formation of the folded wall-and-basin structure province that underlies Expedition Fiord, caused by shortening during the Eurekan Orogeny.

We suggest (Figure 12(b)) that during the Eurekan Orogeny, overpressures were generated under the shifting salt canopy of Jackson and Harrison [6], and successive pulses of hot fluids were expelled, which developed, by hydraulic fracturing, the veins and subparallel zebra rocks of the WGVA in a local tensional (dilatant) environment. It is a characteristic of overpressuring in this environment to reach a point where the effective pressure overtakes the lithostatic pressure and tensional strength of the rocks [107], and a fracture (and a “fracking” earthquake) ensues. Once sufficient volume has been created (dilation) to accommodate the fluid or let it escape, the fluid pressure drops. However, localized pressure drop has little effect on the first-order cause of the tectonic compression, so the pressure builds up again until a new “fracking” event occurs. A sudden pressure drop, combined with decreasing fluid salinity and temperature resulting from mixing of brine and meteoric water, is an optimal mechanism to reduce aqueous CO2 solubility [135, 136]. These processes would precipitate carbonate on the walls of the recently created cavities and fractures followed by quartz precipitates forming the later cavity infilling textures (after carbonate) observed in the veins. However, fluid inclusion observations do not support boiling/CO2 unmixing as a process that took place in the veins. Thus, while pressure fluctuations sufficient to cause fracture formation are mandatory, these pressure fluctuations were not sufficient enough to cause boiling/unmixing. Fluid mixing and cooling were likely important for vein mineral precipitation.

The fluids expelled from the overpressured zones contain small amounts of Cu, Zn, and Pb, which are typical in Mississippi Valley Type (MVT) deposits associated with salt diapirs [115, 137]. The precipitation as sulfides requires the presence of sulfur as dissolved sulfate from the evaporites, and some reduction mechanism to generate hydrogen sulfide. An ideal redox reaction would be the oxidation of organic matter that would be resident in porous sedimentary rocks as petroleum (oil or gas) and the concomitant reduction of dissolved sulfate. Microbial reduction could be a contributing factor, and probably was. The base metal sulfides are generally in veins marginal to the larger carbonate veins, very often in the mafic igneous rocks. The presence of pyrite-marcasite seems to coincide with the presence of mafic igneous rocks rich in Fe (magnetic when fresh). Hence, it is possible that some of the base metals are extracted from the altered diabase and that oxidation of magnetite is accompanied by reduction of sulfate. The fluid expulsion phenomenon was of regional proportions, and if it led to the formation of the WGVA, it is likely to have formed other similar mineralized sites, and this possibility seems confirmed by the presence of gossans in the vicinity of many diapirs in Axel Heiberg Island (e.g., [138]).

The kinship between the WGVA and the modern PSS is further supported by the fluid inclusion data, which suggest a progressive dilution with time from a ~20 wt% NaCl brine down to ~pure water, with a spread of data right across this range. Therefore, we propose that the PSS are exploiting a plumbing system established during the formation of the WGVA (Figure 12(c)).

7. Conclusions

(1)A carbonate vein array with metal sulfides recognized on the western boundary of White Glacier (WGVA) is interpreted to be the remnants of a hydrothermal system developed during the Eurekan Orogeny. Hydraulic fracturing by overpressured fluids led to brecciation of rocks and formation of carbonate veins in dilatant zones peripheral to allochthonous evaporite masses; brecciation of the host rocks may have developed earlier, during displacement of the salt canopies and diapirs (see Section 5.2). Processes envisaged are similar to those leading to the genesis of base metal deposits know as Mississippi Valley Type (MVT) deposits.(2)Initially, the WGVA was formed by hot fluids with a minimum temperature of 150°C to 300°C and having high salinity, in successive pulses with wild pressure changes related to hydraulic fracturing, and later was invaded by dilute fluids of lower temperatures. The depth of emplacement can be estimated to have been a minimum of 0.5 km.(3)The pervasive advection of hot fluids in the Expedition Fiord region explains the anomaly detected by apatite fission track thermochronology which suggested that the rocks now at the surface cooled below 100°C in the Miocene, ca. 20 My later than similar rocks regionally in Axel Heiberg Island and Ellesmere Island. Hydrothermal activity initiated in the Eocene continued for a long time with involvement of cooler, more dilute fluids, extending to this day.(4)Perennial springs (PSS) on Axel Heiberg Island that discharge waters at constant temperatures irrespective of air temperature in an area with ca. 600 m of permafrost are not surficial, periglacial phenomena as previously interpreted, but are instead related to deep plumbing systems established by expulsion of overpressured fluids from under plastically displacing impermeable salt structures during tectonic, compressive deformation and thrust faulting in the Paleogene Eurekan Orogeny.(5)Microbial activity and the involvement of hydrocarbons throughout the history of the WGVA is supported by mineralogy and stable isotope data. If this spring activity has lasted for ca. 10 My as suggested, it could have allowed for isolated evolution of microorganisms since the brines cooled sufficiently to permit their survival. Future work, preferably supported by deep drilling both at the WGVA and PSS, will determine the presence and diversity of microorganisms in these deep subsurface environments.

Data Availability

All data discussed in paper are in 4 tables. Any other data available on request.

Conflicts of Interest

The authors declare that they have no conflicts of interest.


The authors acknowledge support from the Natural Sciences and Engineering Research Council of Canada (NSERC) through grants to M. Zentilli, J. Hanley, and W. Pollard, and an NSERC PDF to C. Omelon. Additional funding for student research was provided by the Department of Indian and Northern Affairs Northern Scientific Training Program and McGill University’s Centre for Climate and Global Change. Field work was supported by the McGill Arctic Research Station (MARS) (c/o Wayne Pollard, without whose support and encouragement this research would have been impossible), the Canadian Space Agency (c/o Gordon Osisnki), the Geological Survey of Canada, and the Polar Continental Shelf Program (c/o Marie-Claude Williamson, Party Chief 2003). We are grateful to Marie-Claude Williamson, with whom the first author first visited the perennial springs, subject of this paper. We thank Alexander (Sandy) Grist for apatite fission-track data and time-temperature modelling, Yawooz Kettanah for fluid inclusion microthermometry from one sample, Dale T. Andersen for stimulating discussions and field photos; John Gosse for advice on the Cenozoic erosion of the study region, Laura Thomson for educating us about glaciers, and Randolph Corney for drafting. We are also grateful to Todd Simpson and Tim Goldhawk of the Western Nanofabrication Facility for technical expertise. This research used resources of the Advanced Photon Source, a U.S. Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Science by Argonne National Laboratory under Contract No. DE-AC02-06CH11357. Data was collected at the X-Ray Science Division at the Advanced Photon Source, Argonne National Laboratory; we thank the Sector 20 beamline staff for their support of this research.