Study Sites

Five of the six retrogressive thaw slump (RTS) sites (NE-14, I-minus 1 and the Itkillik series) are located in the vicinity of the Toolik Field Station (TFS; N68° 38’N, W149° 36’W), which is 720 m above sea level in the foothills province of the Brooks Range, AK (see figure 1, appendix figure A1). Sites were accessed via helicopter from TFS or by foot from the Dalton Highway. Vegetation in this region includes moist acidic tundra (MAT) (tussock sedge, dwarf-shrub, moss tundra; (Walker et al 2002), which is defined by the presence of the tussock-forming sedge Eriophorum vaginatum, moist non-acidic tundra (MNT) (Non-tussock sedge, dwarf-shrub, moss tundra; (Walker et al 2002), and low shrub tundra.

The fourth site (Loon Lake) is located in the vicinity of the Noatak National Preserve (NNP, N67°802–68°839, W155°850–162°8551), which is on the south slope of the Brooks Range in northwestern Alaska (see figure 1). This site was accessed via bush plane from Kotzebue, and then via helicopter from a central field camp. Although MAT and MNT dominates much of the NNP, the tree Picea glauca also occurs at low density (Suarez et al 1999).
To locate sites, we used helicopter over-flights, aerial photographs, Google Earth images (2009) and field surveys. Re-vegetated sites at I-minus 1 were located with LiDAR based on changes in slope topography (Krieger 2012). Once sites were identified, we visually classified disturbance features within sites, henceforth referred to as lobes, as recently disturbed (observed active headwall migration and bare mineral soil exposed due to sediment movement), intermediate aged (mature tall shrubs and some moss cover established) and old (little to no mineral soil exposed, and relatively continuous vegetation cover). Lobes ages (in years after disturbance) were assigned later based on dating of shrubs and soil organic matter (described in an accompanying archived dataset).

Within sites, all lobes were located on the same geologic surface: NE-14 and I-minus 1 were located on the drift of Itkillik phase II (till and ice-contact deposits), Itkillik series on undifferentiated lacustrine deposits, and Loon Lake on Holocene floodplain deposits (alluvium) (Hamilton 2003). Lobes within most sites were close (<500 m) so that the climate and the pool of organisms capable of colonizing were likely similar. The only exception was the Itkillik site with three lobes located along the bank of the Itkillik River and separated by as much as 1 km.

Aspect and topography varied somewhat among lobes within sites. Although the three lobes in NE-14 had similar southern aspects, they differed in slope. The two older sites had ~10% less slope than the active site due to stabilization of the headwalls. The three Itkillik sites had south-eastern aspects according to location on the Itkillik River and slopes were similar among lobes. I-minus 1 site is on the shore of the eponymous lake, and lobes 1-3 had northern aspects, while 4-7 had southern aspects. Slopes were similar among lobes. Finally, all Loon Lake lobes had North-western aspects and similar slopes.
In all sites, undisturbed (control) tundra was randomly selected from the area surrounding the site for comparison. At NE-14, we selected two spatially independent undisturbed areas. At I-minus 1, we selected two undisturbed areas on the south, and one on the north aspect of the lake. For the Itkillik series, we paired an undisturbed area with each disturbed lobe. Only one undisturbed area was sampled in Loon Lake because of helicopter time constraints. In almost all cases, undisturbed sites were gently sloped and had similar aspects to disturbed lobes and generally located upslope from the RTS disturbances.
Within the centre of each disturbed lobe or undisturbed area, we established a 50 m by 4 m belt transect along the contour, where we sampled soils and surveyed vegetation. We sampled a larger area (50 x 10 m along the above transect) for shrub and soil age. If the disturbed lobe was not wide enough to fit a single 50 m transect, two parallel transects were established that covered the same 100 or 500 m2 area. Our primary goal within each lobe was to sample the zone that was not currently affected by active deposition of new material from the headwall or sidewalls or by inundation from the associated lake or river (edge effects). GPS coordinates were recorded for each transect (appendix table A1).

For NE-14, Itkillik and Loon Lake sites, the vegetation of the surrounding undisturbed tundra was classified as MNT: non-tussock sedge, dwarf-shrub, moss tundra, with peaty non-acidic soils (Walker et al. 2002). Frost boils (barren patches of cryoturbated soil) were common. Vegetation at the I-minus 1 site was classified as MAT (Walker et al 2002). Vegetation within disturbed lobes was variable but was dominated by deciduous shrubs (Salix alexensis, S. pulchra, S. glauca, and Betula nana).

Soil C and N pools

Along each transect, we sampled surface soils at 10 m intervals and took additional measurements of organic layer depth at 5 m intervals. At each sample point, we dug a ~30 x 30 cm pit to the organic-mineral interface and removed an organic profile from the exposed wall. Mineral soil was sampled from the organic-mineral interface to 15 cm depth with a 7 cm internal diameter corer. Soils were wrapped in tinfoil to preserve structure and returned on ice to the lab at Toolik Field Station where preliminary processing took place.

At the field station, we sectioned organic soils into depth increments (0-5 cm and 10 cm increments thereafter) with an electric knife. Each increment was weighed and homogenized by hand to remove >2 mm diameter coarse woody debris, roots, rhizomes, and rocks. Gravimetric water content was determined by drying organic soils at 60C for 48 hours and mineral soils at 110C for 48 hours.
We measured soil pH on a 1:1 ratio of air-dried, homogenized soil and DI water. The mixture was allowed to settle for 30 minutes before submerging a calibrated pH electrode (Thermo Orion, Beverly, MA, USA).

To determine bulk soil C and N content, a subsample of the homogenized soil fraction was dried at 60C for 48 hours, ground to a fine powder on a Wiley-mill (Thomas Scientific, Swedesboro, NJ) with a #40 mesh screen, and analysed using an ECS 4010 elemental analyser (Costech Analytical, Valencia, California, USA).

Soil bulk density was calculated for each organic and mineral soil depth increment as the mass (g) per unit volume (cm3) of < 2mm dry soil. Carbon and N pools were calculated for each depth increment as the element concentration times the bulk density scaled to a meter squared. The total organic layer pool is the sum of all depth increments within a profile. Organic layer bulk density was calculated as an average for each profile by dividing the summed soil mass by the summed volume. Total layer %C and %N were similarly calculated for each profile by dividing the summed C or N pool by the summed soil mass. Total layer pH was calculated for each profile by (1) multiplying hydrogen ion concentration in each layer by the mass of soil in the layer, (2) summing the layers, and (3) dividing the sum by the total soil mass. Statistics were performed on hydrogen ion concentrations and back-transformed to pH for reporting.