The role of periglacial processes on soil carbon distribution is examined at a High Arctic site in northwest Greenland. A 16-m trench dug across a series of nonsorted stripes at Thule Air Base revealed sand-rich wedges underlying striped, vegetated troughs, and organic-rich soil horizons buried at depth. The site has sparse prostrate vegetation and is estimated to contain 9.4 kg/m2 of soil organic carbon (SOC) in the active layer. The distribution of carbon is variable with nearly half (49%) stored in the sand wedges, which only account for 10% of the trench area. Additionally, 62% of the total SOC was found below 25 cm, highlighting the significant role of cryoturbation and physical redistribution of carbon in permafrost-affected soils. Carbon in the active sand-rich wedges dates from modern at the surface (65 ± 35 radiocarbon years) to 2695 ± 40 radiocarbon years at depth, and carbon turnover time appears to be ∼450 years. Buried organic horizons found at 50–70 cm depth have radiocarbon ages of 27,480–31,900 BP. A conceptual model is proposed in which the active sand wedges have developed in an approximately 30 ka surface containing buried soils preserved in permafrost or under a cold-based glacier. As the ice retreated and soils warmed, soil development and active cryoturbation resumed forming nonsorted stripes in the modern surface.
 Patterned ground forms in arctic permafrost terrain as a result of cryoturbation; the mixing, heaving, and churning of soil that occurs during freeze-thaw cycles [Washburn, 1980]. These cryogenic processes form stripes, circles, polygons, and nets with and without visible surface textural sorting [Washburn, 1980]. Subsurface expressions of cryoturbation include: convoluted or disrupted soil horizons, oriented rocks, and accumulations of organic matter near the top of the permafrost table [Bockheim and Tarnocai, 1998; Tedrow, 1965; Vandenberghe, 1988; Washburn, 1980].
 Nonsorted stripes are one of the more common forms of patterned ground on sloping surfaces of the High Arctic [French, 1976; Washburn, 1980]. Washburn  defines nonsorted stripes as “patterned ground with a striped pattern and a nonsorted appearance due to parallel lines of vegetation covered ground and intervening strips of relatively bare ground oriented down the steepest available slope.” They are distinguished from sorted stripes in that they are not texturally sorted on the surface and have stripes of vegetation associated with linear depressions [Washburn, 1956]. These features have been occasionally referred to as solifluction stripes because of their downslope movement [French, 1974; Washburn, 1947]; however, the term has largely gone unused and solifluction is generally used to describe lobate features. Both the vegetated and nonvegetated portions of nonsorted stripes can be equal in width but typically reported vegetated portions are 0.2–0.6 m with wider nonvegetated portions ranging from 0.5 to 4 m [French, 1974; Poser, 1931; Washburn, 1947]. Nonsorted stripes can assume large- (as above) or small-scale (nonvegetated stripes of ∼0.16 m) forms and occur on slopes of 3–20° [Washburn, 1980].
 The long-term stability of SOC can be estimated from the age of SOC. One approach to evaluate the current decomposition rates is to estimate SOC turnover based on age and input rates of carbon [Trumbore and Harden, 1997]. This approach is utilized here, however, it is limited to providing only bulk turnover rates and cannot account for multiple soil carbon components with various degrees of stability as it is done in more detailed ecosystem models [Clein et al., 2000; Jenkinson et al., 1991; Jones et al., 2005; McGuire et al., 1995].
 The present paper investigates the importance of patterned ground in moving SOC to depth, and the stability of buried SOC in the High Arctic. It presents a case study of the age, amount, and spatial distribution of SOC associated with a series of nonsorted stripes in northwest Greenland. The study focuses on soil samples collected from a 16 m long and 0.7 m deep trench cut transverse to a series of four nonsorted stripes. The trench provides an extraordinary view of subsurface features across several stripes and gives insights into processes governing the distribution of SOC.
2.1. Site Description
 This study was conducted near Thule Air Base in northwest Greenland (76°N, 68°W) in the summer of 2004. The base is located on an 800 km2 ice-free peninsula (Pittuffik) bounded to the east by the Støre Landgletscher (which is connected to the Greenland Ice Sheet) and Baffin Bay to the south and west (Figure 1). Mean annual air temperature (1978–2003) of the area is −11.9°C with growing season (June, July, and August) temperatures averaging 3.8°C, and mean annual precipitation (1978–2003) is 118 mm, with the growing season averaging 50 mm (Thule Air Base, unpublished data). According to the Circumpolar Arctic Vegetation Map (CAVM) classification, this region would be in bioclimatic zone C [CAVM Team, 2003]. The surficial geology of the peninsula is dominated by glacial drift of mixed lithology with outcrops of sedimentary formations (sandstone, dolostone, limestone, basalt sills) in the north and crystalline basement rock (pink and gray banded gneiss) in the south [Dawes, 1976].
 The trench site is located south of the air base at the crest of a 200 m high bedrock “island” known as South Mountain or Akînarssuaq (76.51°N, 68.67°W) (Figure 1). The underlying geology is a Proterozoic sedimentary formation composed primarily of dolostone, siltstone, and sandstone covered by a thin veneer of glacial drift [Davies et al., 1963; Dawes, 1976]. Nonsorted stripes appear in soil formed from a mixture of glacial drift and bedrock on a south-facing slope of 3%. The stripes are composed of alternating near-barren soil ridges (1–3 m wide) and narrow vegetated troughs (0.5 m wide) (Figure 2). Although the barren ridges contain a large amount of surface gravels and cobbles, there is no obvious sorting on the surface and thus by definition are classified as nonsorted [Washburn, 1956]. A slight linear depression often delineates the centerline of the sparsely vegetated ridges (Figures 2 and 3) . Vegetation in the troughs is dominated by Dryas integrefolia and lesser amounts of Salix arctica and Saxifraga oppositifolia. The vegetation cover of the site is approximately 33%, with the majority of the vegetation confined to the troughs.
2.2. Trench Digging and Soil Mapping
 A 16 m long trench perpendicular to the nonsorted stripes was excavated with a backhoe on June 30, 2004. The trench bisected four vegetated troughs and five near-barren ridges, providing significant replication and a more comprehensive view of the stripes than a single soil pit. The thaw depth in late June was approximately 50 cm; therefore the pit was left open for 1–2 days to allow thawing to greater depths. An additional 20 cm of soil was hand excavated to a maximum trench depth of 70–74 cm. As this trench was dug early in the summer thaw season, it likely did not reach the base of the active layer; seasonal thaw depths at nearby research sites can exceed 120 cm.
 A leveled line was established along the northern (upslope) face of the trench, approximately 20 cm above the surface. This line was used as a reference for mapping the surface topography of the stripes and the soil horizons (Figure 3). Soil horizons were described using diagnostic soil horizon designations (i.e., O, A, C, etc.) and samples (∼1 kg) of each horizon were collected. The profile sketch was later digitized and the area of each horizon calculated using ESRI GIS and mapping software. According to the World Reference Base for Soil Resources (WRB) classification system, the soil of the stripe complex is classified as a Turbic Cryosol [FAO, 2006].
2.3. Soils Analysis
 Soil samples were dried and sieved (2 mm mesh) for 20 min using an electric sieve shaker (Model 150, Derrick Manufacturing, Buffalo, NY). The <2 mm particle size distribution (fine fraction) was determined by laser diffraction (Beckman-Coulter LS 13–320) at the University of Washington. In addition, >10 kg soil samples were collected, dried, and sieved to determine the >2 mm fraction (coarse fraction). Soil pH was determined using a 5:1 ratio of dry soil to dionized water and measured with a glass electrode pH meter (WTW 340i multimeter). Soil bulk density was determined using soil moisture containers for fine-textured soils and large bulk samples for gravelly soils (refer to Horwath  for additional methodology details).
2.4. Carbon Analysis and Calculation
 Organic carbon was analyzed using a Perkin Elmer 2400 carbon, hydrogen, and nitrogen (CHN) elemental analyzer at the University of Washington. A 2 g subsample of the <2 mm fraction was dried at 60°C for 24 h and hand ground with a mortar and pestle. If carbonates were present, samples were pre-treated with a 1:1 HCl solution. Using an eyedropper, HCl was applied to a small amount of sample in a ceramic dish. After bubbling ceased and carbonates were removed, the sample was oven-dried and stored in a desiccator until analysis. Approximately 30–50 mg of sample was placed in a tin capsule and ignited in the elemental analyzer. A standard or duplicate check was run every 10 samples.
 To calculate the total SOC of the trench, the area of each horizon was first calculated from digitized maps of the trench wall. The SOC for each horizon was determined by multiplying the carbon content of the <2 mm fraction (kg/kg) by the mass fraction of <2 mm content, the bulk density of the whole soil (kg/m3), the fractional area of each horizon, and the total pit depth (m). The carbon content of all horizons was then summed and normalized to 1 m2 surface area to obtain the total trench carbon content.
2.5. 14C Analysis
 For 14C analysis, 200–300 g soil samples were collected from the trench using a clean spatula and Nitex gloves, and kept frozen (−20°C) in sterile Whirlpack-brand bags until time of processing. Frozen samples were thawed and a 1–3 g subsample was removed and dried at 105°C for 24 h. The dry samples were sieved (0.5 mm mesh) for 20 min using an electric sieve shaker (Model 150, Derrick Manufacturing, Buffalo, NY). The <0.5 mm fraction was examined under a binocular microscope to remove any visible modern roots and stems, as they could bias the dating toward a younger age. After each sample was free of obvious modern roots, it was ground with a mortar and pestle that was carefully cleaned between samples. An aliquot of each sample was used for carbon analysis (by CHN analyzer) to determine the amount of sample necessary for dating. To remove carbonates, 10 ml of 0.1 N HCl was added to the sample (in a glass centrifuge tube), placed on a vortex mixer for 1 min, and then heated at 90°C for 15 min to hasten carbonate removal. Samples were centrifuged for 10 min at 3000 rpm followed by aspiration of the supernatant and drying overnight at 90°C. Samples were transferred to glass vials and shipped to the Lawrence Livermore National Laboratory for Accelerator Mass Spectrometry (AMS) analysis. To ensure quality control one duplicate sample and two standards were prepared.
3. Results and Discussion
3.1. Soils Results
 Ten soil horizons were identified in the trench (descriptions in Figure 3 and Table 1). The most distinct features are well-defined wedges of sand-rich soil that taper downward directly beneath the vegetated troughs (A2). These dominantly fine-textured horizons are surrounded by distinct borders of gravels (C3). Soil wedges beneath vegetated stripes in Svalbard and northern Canada were noted in cross-section by Poser  and Washburn , but not to the depth and width observed here.
Table 1. General Soil Descriptions and Soil Organic Carbon (SOC) Estimates for Nine of the Ten Soil Horizons Observed in the Trencha
X Coord (m)
Y Coord (m)
% >2 mm
% <2 mm
% Organic Carbon
Whole Soil BD
Horizon Area (m2)
Organic C Content (kg/m2)
No data were collected for A1 horizon. Note that y coordinates indicated depth from surface.
 Another prominent subsurface feature of the nonsorted stripe complex is a dark brown, discontinuous horizon of organic-rich soil generally buried near the base of the trench excavation (A4@ and A5@) (@ = WRB cryoturbation identifier. (Although this is a buried horizon, the use of subordinate “b” designation (buried) is not allowed for cryoturbated horizons in the WRB classification system.) This horizon is most often observed as isolated “pockets” of organic-rich soil near the base of the active layer beneath the near barren ridges, but in a few circumstances is found connected to the surface (Figure 3). Some pockets exhibit wave-like structures suggestive of deformation (Figure 2). The presence of organic pockets beneath barren stripes was also observed by Walmsley and Lavkulich  in the Mackenzie Valley region of northern Canada, although not to the size and extent of those observed here. Other organic-rich horizons (A3 and O) are found near the surface associated with vegetated troughs and centerline depressions (Figure 3).
 The other horizons are classified as mineral soils that meet the criteria for C horizon; they are light in color and have low carbon contents. Portions of the barren ridge surfaces are characterized by a salt-rich, carbonate cemented C horizon (Ckm) that can extend from the surface to depths of 30 cm. The cementing is observed only in the surface horizons of the barren ridges, and likely results from water evaporation and the concentration of soluble ions. Other C horizons (C2 and C4) appear to be derived from glacial drift of mixed lithology and extend from the surface to the base of the trench.
 Textures of the soil horizons range from loams and sandy loams to loamy sands, with all horizons having low clay content (2.5–11.4%) (Table 1 and Figure 4a). All C horizons (generally located beneath barren ridges) have similar texture, with 50–60% sand, 30–40% silt, and approximately 10% clay. Sand wedges (A2) and buried organic-rich horizons (A4@ and A5@) have similar soil textures, with compositions of 75–80% sand, 17–20% silt, and <5% clay. All C horizons have a higher silt content than sand wedges or buried organic horizons (17–20%), causing them to be more susceptible to frost heaving [Penner, 1968]. The mechanism for the accumulation of sand in the wedges is unclear, but potential mechanisms include washing out of fine fraction material (i.e., silts and clays) and vegetation trapping of eolian sands [Washburn, 1980]. This textural difference may contribute to the surface stripe formation by stabilizing the sand wedges beneath the vegetated troughs relative to the more frost-susceptible soils beneath the barren ridges. The clear association of vegetation with the sand wedges suggests a symbiotic relationship where the sand wedges provide stable substrate for vegetation growth, and the vegetation traps saltating grains.
 While the surface of the stripe complex shows no obvious sorting, considerable sorting is evident below the surface. Sand wedges have relatively low gravel content with compositions of 80% fine and 20% coarse fraction. In almost all instances, sand wedges are bordered on either side by a gravel lens (C3) containing 80% coarse and 20% fine fraction (data from large bulk samples) (Figures 3 and 4b). Gravel in this zone is tightly packed and often in point to point contact. Horizons located beneath the barren ridges (C2, C4, etc.) do not exhibit the distinct sorting observed in the sand wedges, but average 30–40% coarse fraction and 60–70% fine fraction. This relatively low gravel content may result from the comparatively high silt content of the C horizons and the ability to more easily heave gravels to the surface, thus slowly removing them from the horizon. Although the fine fraction in sand wedges and buried organic horizons is similar, the coarse fraction differs greatly. Buried organic horizons have an average of 67% coarse fraction, while sand wedges are relatively gravel-free with an average of 4.5% coarse fraction (data from smaller horizon samples) (Table 1 and Figure 4b).
 Dry bulk densities for the <2 mm fraction range from 0.9 g/cm3 in the O horizon to 1.6 g/cm3 in the Ckm horizon. When the gravels are included in the bulk density calculation (i.e., “whole soil” bulk density), values increase and range from 1.1 g/cm3 (A2) to 1.9 g/cm3 in A5@ (Table 1). While sand wedges and buried organic horizons have similar bulk densities for the <2 mm fraction, the increased gravel content in the buried organic horizons results in higher whole soil bulk density values.
 Soil pH values range from 7.4 in A5@ horizons to 9.2 in carbonate-cemented C horizons (Ckm) on the surface of barren ridges (Table 1). The more neutral pH values reflect the nature of the mixed lithology that comprises the glacial drift, while the high values associated with surface cementing reflect the influence of evaporation and the concentration of salts in this region. Formation of salt crusts during dry weather periods has been observed in many locations across the study area. Analyses of such salt precipitates collected at the base of South Mountain indicate high sodium concentrations (B. Hagedorn, unpublished data) suggesting a marine source (assuming that sodium-rich salts mix with carbonate-rich soil water to form bicarbonate salts thus raising soil pH to values >9). C horizons below the surface (e.g., C2, C3, and C4) have pH values between 7.8 to 8.8 reflecting the carbonate lithology and possibly salt accumulation. The sand wedges (A2) and buried organic horizons (A4@ and A5@) have similar pH values (7.4 to 7.8) also reflecting the presence of carbonate bedrock.
3.2. Organic Carbon Analysis and Distribution
 Organic carbon values range from 0.14% (by weight) in Ckm horizons to 10% in the surface of the sand wedges (A2) (Tables 1 and 2). SOC in the sand wedges (A2) generally decreases from the surface to depth, and remains relatively high for arctic soils, ranging from 5.8 to 7.5% at 25 cm to 2.2% at 50 cm. SOC in buried organic-rich horizons (A4@ and A5@) is generally uniform and less than the sand wedges, but 4.1% (50 cm) and 1.6% (70 cm) SOC is still a considerable amount at that depth. Both the buried organic-rich horizons and sand wedges stand out as carbon-rich pockets contrasting with the surrounding C horizons that all contain less than 1% SOC.
Table 2. Organic Carbon Data and Radiocarbon Ages for Ten Soil Samples (One Sample Ran in Duplicate) Analyzed by AMSa
Lawrence Livermore Number
X Coord (m)
Y Coord (m)
% Organic C by CHN
Calendar Years b.p.
Note that these samples were collected separately from the samples displayed in Table 1. Comments: d13C values are the assumed values according to Stuiver and Polach . The quoted age is in radiocarbon years using the Libby half-life of 5568 years and following the conventions of Stuiver and Polach . Radiocarbon concentration is given as fraction modern, D14C, and conventional radiocarbon age. Sample preparation backgrounds have been subtracted, based on measurements of samples of 14C-free coal. Backgrounds were scaled relative to sample size. Calendar age calibration determined with the CALIB 5.0 software program and are reported as 1 sigma. The four oldest samples were beyond the range of calibration. Asterisk indicates highest probabilty of date ranges under 1 sigma.
sand wedge - A2
sand wedge - A2
buried organics A5@
buried organics A5@
buried organics A4@
buried organics A4@
sand wedge - A2
0, 34–72* 116–135, 226–252
sand wedge - A2
sand wedge - A2
 The total SOC content for all the horizons along the 16-m trench averages 9.4 kg/m2 (Table 1), however, SOC is highly variable throughout the nonsorted stripe complex. Nearly half of the total SOC (49%) is stored in the sand wedges, which account for only 10% of the trench area. As the vegetation in the troughs is the most likely source of carbon to the sand wedge, this distribution is not surprising. SOC appears to be incorporated from surface vegetation and remains localized in the sand wedge. Carbon-rich, discontinuous horizons (A4@ and A5@) found beneath the barren ridges account for 19% of the total carbon. Combined with the carbon content of the sand wedges, these carbon-rich regions contain 68% of all the SOC and yet cover only 26% of the trench area. The remainder of the SOC is located in C horizons which, although low in organic carbon, cover the largest extent of the profile.
 Over 62% of the SOC measured in the trench excavation is located below 25 cm depth. As noted by Horwath , previous soil studies by Bliss and Matveyeva  likely underestimated High Arctic SOC in certain vegetation communities by more than an order of magnitude, since soil was only sampled to a depth of 20–25 cm. The present data show that sampling of only the upper 25 cm would have underestimated the carbon pool by over 50% and that there is likely more SOC buried in the active layer below 75 cm (which was still frozen at the time of sampling), and also in the permafrost (>120 cm).
3.3. 14C Analysis
 AMS dating of bulk soil suggests that two distinct pools of carbon exist beneath the nonsorted stripes of South Mountain (Akînarssuaq). Although it was anticipated that buried organic-rich horizons (A4@ and A5@) would be both physically and spatially connected to the sand wedges (A2), the data show that the two horizons represent two pools of carbon formed at different times.
 Organic carbon in the sand wedges (n = 6) is relatively young and increases in age from 65 ± 35 radiocarbon years in the upper 5 cm to 2,695 ± 45 radiocarbon years at 60 cm depth (Figure 5 and Table 2). The range of dates is similar in the two sand wedges sampled, and reflects the input of modern carbon from the abundant vegetation in the troughs and is likely labile. Although modern roots were removed from each sample, the possibility exists that a few very small roots may have remained. If so, the bulk dates would possibly reflect a mixture of older subsurface carbon mixed with a small percentage of modern material; a point to be discussed later in the paper.
 Four dates obtained from two discontinuous organic-rich horizons located beneath the barren ridges range from 27,480 ± 320 to 31,900 ± 540 radiocarbon years (Figure 5 and Table 2). These dates are older than previous reports of organic carbon ages in the Arctic (1,695–11,000 years BP) (summaries in Everett et al. ). The carbon buried beneath the ridges is distinctly older than the carbon in the sand wedges. The buried soil horizon ages are similar within and between each other, suggesting that both “pockets” have a similar history. The age and uniformity of dates suggests that (1) the buried soil has most likely been “locked” away from microbial decomposition for a long time or (2) the carbon is recalcitrant. The most likely mechanisms of preserving carbon at depth include: carbon being frozen into aggrading permafrost as temperatures decreased [Dyke and Zoltai, 1980], burial by periglacial processes including solifluction or mass wasting [Washburn, 1980; Worsley and Harris, 1974; Zoltai et al., 1978], or burial by deposition of glacial drift [Zoltai et al., 1978]. Most soils in the Arctic have formed since the last glaciation. Some examples of soil ages in the Arctic include: 2,450–9,360 years BP in basal peat on Devon Island [Barr, 1971], 2,000–11,000 years BP in buried organic remains at Point Barrow, Alaska [Brown, 1965], 3,340–9,480 years BP in High Arctic earth hummocks [Zoltai et al., 1978], 1,035–7,370 years BP buried in mudboils of northern Canada [Dyke and Zoltai, 1980], and 420–6050 years BP in soils of northern Alaska [Marion and Oechel, 1993]. Continuously cold and wet conditions resulting in anaerobic soil conditions could also slow (but probably not stop) carbon decomposition, but no gley (reducing) conditions were observed at depth, and the sloping terrain and coarse texture of the gravels bordering the sand wedges favors effective drainage of the active layer. Thus the absence of gley in these water saturated soils suggests that there is little biological activity, and supports a scenario whereby the oldest carbon pool was frozen.
 Assuming that the carbon content across all sand wedges (A2) and buried organic-rich horizons (A4@, and A5@) is relatively homogenous, young carbon in the sand wedges is 4.6 kg/m2 SOC (49% of the total SOC), while the older carbon pool is 1.8 kg/m2 (19% of the total SOC). Although the concentration of young carbon is larger, the pool of older carbon may be a significant component as exemplified by Winston et al.  and Schimel et al.  who suggested that older pools of carbon in the Canadian boreal forests and Alaska tundra soils, respectively, may significantly contribute to winter CO2 flux.
 The interpretation of radiocarbon dated soil organic matter is not trivial. One issue is that soils are continually developing and new SOC is being added to the system by surface inputs, root decomposition at depth, or by leaching of DOC and transport to depth. This leads to 14C dates that are always younger than the age of the oldest carbon component [Wang et al., 1996]. Because SOC is composed of various fractions of organic matter (e.g., humin, humic and fulvic acids, lignins, etc.) which may vary in age and stability, a bulk soil age will always necessarily provide an overall average age of the SOC [Chichagova and Cherkinsky, 1993]. In arctic soils, cryoturbation is an additional confounding factor since it disrupts the stratigraphy and thereby complicates the interpretation of soil ages.
 Given these complexities, it is probable that the average ages obtained for the sand wedges are younger than the actual age of the soil. This issue is less likely for the buried organic pocket because sustained input of “young” carbon is unlikely at depth, especially if the soil is frozen [Orlova and Panychev, 1993]. Although no major root systems were observed in the buried organic deposits, small vestiges of roots that often accumulate on the top of the permafrost table could be incorporated in the dated organic deposit, which would lead to younger dates. Contamination by older carbon components, such as coal, is not likely in the buried organic deposits in that there are no coal-containing bedrock types in the region [Davies et al., 1963]. One might argue that both the SOC in sand wedges and buried pockets were originally the same, with the former largely contaminated by modern carbon from overlying vegetation. This seems unlikely though because a mixture between modern carbon and the old carbon pool (∼31 ka) suggests that the young carbon pool (∼1.1 ka) would be a mixture of ∼76% of modern carbon and 24% of the old carbon pool (∼31 ka). (Using the equation for calculating radiocarbon ages: age = 8033*ln(1+D14C/1000) [Stuiver and Polach, 1977], and assuming the oldest age has −980 D14C (31 ka) and modern carbon has 140 D14C, suggests that a carbon pool with age of 1.1 ka would be a mixture between ∼76% of modern carbon and ∼24% of old carbon.)
 The old age and reproducibility between the separate pockets suggests that mixing with young carbon did not influence the ages significantly as mixing with various amounts of “young” organic matter would tend to lead to a range of ages. However, dating macrofossils in these organic deposits (if present) would better constrain the average dates and elucidate the potential influence of younger carbon. Although caution is warranted in the interpretation of bulk soil ages, it is clear that there are two distinct pools of carbon in this patterned ground system of northwest Greenland. Whether these drastically differing pools are found commonly in High Arctic soils remains to be determined.
3.4. Dynamics of SOC and Carbon Turnover
 Radiocarbon ages provide a tool to investigate the carbon dynamics in this patterned ground system. Following Harden et al.  and Trumbore and Harden , vertical profiles of SOC and associated radiocarbon dates in the sand wedges were used to determine approximate turnover and carbon input rates. The net change of carbon in soil is a function of carbon input rate (I, mass/area/time) and carbon decay (kC) for the soil volume, as described in equation (1). Here k is first order rate constant for loss (1/time) and C is the carbon content (mass/area). Integrating equation (1) yields equation (2) and, if I and k are known, the carbon inventory of the soil at various times can be calculated. After an initial buildup of carbon in the soil, a steady state is established (t → ∞) with the carbon input equaling the carbon decay (C = I/k). The inverse of k is effectively the average residence time of carbon (also-called turnover time) in the soil unit [Albarède, 1996]. As shown by Trumbore and Harden  this can be used to model carbon dynamics (I,k) from soils where age of organic carbon is known.
 Use of the model in the trench soils is similar to that used by Trumbore and Harden  except that here it is assumed that the sand wedges grow upwards due to soil accumulation in the wedges over time. It is also assumed that the rate of carbon input is constant, the soil carbon pool is homogenous, and climatic conditions have remained relatively constant.
 The present study uses the best fit of 14C age and carbon inventory of sand wedge 2 (refer to Figure 5 for sand wedge labels). For calculating the carbon inventory it was assumed that the SOC is constant to a point equidistant above and below sampling point. For fitting the curve, we used the add-in solver provided in Excel®. For this I and k were varied until the best fit was found (Table 3 and Figure 6). The values I and k were constrained as I ≤ 100 gC/m2 and k > 0. The best fit is reached with an annual carbon input rate of 100 g C/m2/year and a turnover rate of about 452 years. The input rate is substantially higher than reported by Harden et al.  for Cryosol soils on Baffin Island (4.3 g C/m2/year) but could possibly approach this rate if the autochthonous transport of organic material by wind and water into the troughs is accounted for. Restricting the input rate to ≤50 gC/m2 yields the turnover rate of ∼1200 years, but the fit is poorer than with the higher input rate of 100 gC/m2. This preliminary calculation is based on three data points in a single sand wedge and C14 analysis with a more detailed sampling approach can verify and improve these first results.
Table 3. Data Used in the Sand Wedge Soil Turnover Rate Calculationa
Sampling Depth (cm)
Thickness of Layer (cm)
Fraction of Horizon
C per Layer (gC/m2)
Cumulative Inventory (gC/m2)
Input Rate (gC/m2/year)
Although the surface O horizon was not sampled in wedge 1, it is present. The percent carbon and bulk density of the O horizon in wedge 2 was applied to wedge 1 (as an approximation) in order to more completely account for the total SOC storage.
 The carbon turnover rate of 450 years (Table 3) is over twice as fast as reported by Harden et al.  for Cryosols of Baffin Island (1100 years) but is slower than for temperate soils [Harden et al., 1997]. While the turnover rate is slow, buried carbon should decay within a few thousand years. Assuming that the old carbon found close to the permafrost table in the A4@ and A5@ horizon (27,480–31,900 radiocarbon years) has equivalent compounds and lability, the turnover rates suggest that the old carbon must have been preserved to survive such long periods. The most likely scenario to preserve SOC for this long is storage in permafrost at subzero temperatures. Warming the soil and deepening of permafrost table therefore could lead to fast turnover of the old carbon and its ultimate release to atmosphere.
3.5. Interpretation of Two-Pool SOC Origin
 The origin of the dichotomy in soil carbon ages merits attention. A polygenetic origin is proposed here whereby the currently active sand wedges have developed in an older surface that was buried and preserved by glacial drift. The current surface is shaped by periglacial processes responsible for the active nonsorted stripes that characterize this terrain. Based on the radiocarbon dates, carbon dynamics, and current arctic geomorphic processes, three formative phases are suggested:
 Phase 1: Prior to 32 ka the Thule region was likely ice-free, and possibly covered with vegetation and periglacial features similar to those present today. This is in agreement with the Quaternary glacial history of the region proposed by Kelly et al.  and Kronborg et al.  and with the regional climate of the North Atlantic as indicated by oxygen isotope data obtained from the GISP2 ice core at the summit of the Greenland Ice Sheet [Stuiver and Grootes, 2000]. These isotope data suggest annual temperature at 200 m asl of approximately −16.1°C; 4.2° colder than the modern mean annual temperature of −11.9°C recorded at Thule Air Base (Figure 7). (This value was obtained using the calculation of 0.33 per mil change in δ18O per 1°C, derived by Cuffey et al.  from borehole temperature data of the Greenland Ice Sheet. The dry adiabatic lapse rate of 8°C/km was used to translate temperature from the summit of the Greenland Ice Sheet (3200 m asl) to the approximate modern elevation of South Mountain (200 m asl). This does not take into account any glacial rebound or depression elevation changes at 30–32 ka.) Similar vegetation is noted in other regions of the High Arctic, e.g., Alert and Eureka, Canada where mean annual temperatures are −18 and −20°C, respectively. The soils most likely resembled those of today, with a moderate accumulation of organics and discontinuous A or O horizons and being impacted by cryoturbation.
 Phase 2: At approximately 32 ka, ice advanced from the north, extending south to the crest of South Mountain and burying the surface with a thin glacial drift. Davies et al.  note the presence of an ice front boundary on South Mountain. They note that marginal channel perimeters northeast of South Mountain are concentric northward, while channels south of South Mountain are concentric in a southward direction, suggestive of an ice boundary. By radiocarbon dating marine shells in the glacial drift of Saunders Island (Figure 1), Davies et al.  estimated this glacial advance to be 32 ka. Similarly, using dated marine shells from cores in Wolstenholme Fjord, Kelly et al.  propose that the advance occurred around 24 ka. The burial of the soils in this age range would account for the older carbon, which ranges from 27.5 to 32 ka; and the colder conditions at this time would help account for the preservation of the carbon. The oxygen isotope data from the GISP2 ice core show that between 32 and 30 ka, δ18O dropped to some of the lowest values attained in the ice core record, corresponding to mean annual temperature of approximately −28.2°C, which is 12.1°C colder than during Phase 1, and 16.3°C colder than the modern mean annual temperature at Thule Air Base (Figure 7). The Phase 1 soil would now be preserved under the frozen fringe of the ice sheet and isolated from new carbon input. The temperature record from the GISP 2 ice core (Figure 7 and Cuffey et al. ) suggests that pre-Holocene temperatures were likely cold enough to lead to shallow active layer depths and to a lack of cryoturbation at depth (e.g., 70 cm).
 Phase 3: (Holocene) As climatic conditions slowly warm to those of modern day, the reestablishment of a vegetated surface and periglacial patterning of the ground resume. Among the many processes likely to be involved in the formation and maintenance of nonsorted stripes, as well as other forms of patterned ground [e.g., Washburn, 1980], contraction cracking and trapping of eolian sands may account for the observed sand wedges. Gravel may also accumulate at the soil surface and edge of the wedges below the surface because isotherms in the soil are not planar and hence the upfreezing is not simply normal to the surface. Instead, the freezing isotherm will tend to dip down over the wedges, parallel to the local ground surface and hence some subsurface convergence of gravel is expected due to the wedge-ward horizontal component of upfreezing. Although the wind-blown origin of the sand in the wedges is not certain, the texture of the wedges is sand with a mean diameter of 154 μm in a size range that saltating grains could readily be transported by wind (Figure 8) [Bagnold, 1954; Mabbutt, 1977].
 These sand wedges are similar in terms of texture and general shape to those associated with thermal contraction cracks in the Dry Valleys of Antarctica that seasonally open and fill with eolian sand [Sletten et al., 2003]. They also bear resemblance to the ice wedges associated with thermal contraction cracking patterns in the Arctic as observed by Lachenbruch . Although similarities exist in the form of the wedges under study, their ∼2 m spacing contrasts with the 10 to 30 m length scale characteristics of thermal contraction patterns.
 The progressive growth of the sand wedges pushes material aside including the buried organic deposits. For example, the wave-like structures in Figure 2 suggest movement of material away from the sand wedge. The progressive growth and widening of the sand wedges also tends to mound the surface of both sides of the sand wedge, which may account for or contribute to the broad near-barren ridges bordering the wedges; the previously noted sparsely vegetated depression marking the center of the ridges could then be in large part the low spot between adjacent “shoulders” (Figure 3). As the sand and gravel wedges are less susceptible to frost heave and provide a stable growth medium, plant species likely accumulate in troughs overlying the sand wedges and contribute organic carbon by litter fall and root decomposition.
 Much like other forms of highly elongated patterned ground, the nonsorted stripes in the study area reflect active slope processes such as solifluction. Numerical experiments of Kessler and Werner  provide valuable mechanistic insights into the development of sorted patterned ground on hillslopes, which consist primarily of stripes aligned along the steepest descent. Although the nonsorted stripe system at South Mountain is not analogous to that of Kessler and Werner, their model provides insights into the influence of downslope movement on patterned ground, and shows that equidimensional patterned ground features naturally transition to linear stripe-like features as the hillslope angle increases. Preliminary high-resolution GPS data suggest rates of slope movement of approximately 1 cm/year on South Mountain, which is similar to downslope movement from similar arctic systems (2.6 cm/year for nonsorted stripes in northern Canada [French, 1974] and 1–3 cm/year in Svalbard [Jahn, 1960]).
 Nonsorted stripes are relatively common landscape features in the High Arctic where SOC storage has been underestimated substantially and the potential CO2 contribution of deep SOC to atmospheric carbon may also be underestimated. Nonsorted stripes reveal distinct wedges of sand-rich soil horizons below vegetated troughs with intervening barren ridges containing gravelly silt above buried organic-rich horizons. The total SOC storage determined over the length of the trench is 9.4 kg C/m2. Nearly half of the carbon (49%) is stored in the sand wedges (A2), which only account for 10% of the trench area. The carbon stored below 25 cm represents 62% of the total carbon storage. The carbon content and distribution is similar to that determined for a range of similar soils in the Thule region that averaged 6.9 kg C/m2 with an average of 46% stored below 25 cm [Horwath, 2007]. Assuming that the carbon content across all sand wedges and buried organic-rich horizons is relatively homogenous, young carbon in the sand wedges is 4.6 kg/m2 SOC (49% of the total SOC), while the older carbon pool is 1.8 kg/m2 (19% of the total SOC).
 Radiocarbon dating of bulk soils revealed two distinct pools of carbon: a younger pool with modern (65 ± 35) to 2,695 (±40) radiocarbon years that is confined to the sand wedges directly beneath the vegetated troughs, and an older “preserved” pool located largely at depth. The pockets of older SOC buried at 50–70 cm have 14C ages of 27,480–31,900 radiocarbon years. The 14C dates and carbon inventory in the sand wedges provide a modeled turnover rate of approximately 450 years. These rates suggest that the old carbon pockets has been preserved, most likely being frozen in the permafrost or buried by the cold edge of a prior ice sheet.
 Based on these findings, a three-phase conceptual model is proposed for the development of a two-pool SOC system: (1) about 30,000 years ago the site was shaped by periglacial processes similar to those that are currently active; (2) the surface and associated SOC were buried by glacial drift (supported by regional temperature data from the GISP2 ice core and glacial geomorphology) as the site was overridden by the terminus of the nearby ice cap; and (3) the ice retreated, vegetation established on the surface again, and active cryoturbation resumed creating the currently active nonsorted stripes.
 Nonsorted stripes appear to foster carbon entrainment at depth. The older carbon pool presented here may be unique to the South Mountain site, but additional research is needed to provide accurate estimates of SOC content and age distribution across a range of High Arctic environments.
 Funding for this research was provided by the National Science Foundation (OPP-0221606); University of Washington Department of Earth and Space Sciences; Geological Society of America (7691–04); Association of American Geographers (AAG) Dissertation Research Award; AAG – Geomorphology Specialty Group Dissertation Research Award; Arctic Institute of North America Grant-in-Aid Award. Logistics and field support provided by VECO Polar Resources, and the U.S. Air Force. We thank J. Welker for informative discussions and project leadership, H. Kokorowski and A. Vigna for field assistance, and T. Brown for his expertise and advice in carbon dating.