A loader tractor with a tooth bucket was used to create grooves across the disturbed area which had been heavily compacted by machinery. Grooves were created by dragging the bucket perpendicular to the direction of prevailing winds, creating ridges approximately 10 - 15 cm deep, and 20 cm apart. In 2009, it was suggested by an on-site remediation design engineer that a section of the disturbed area be amended with an Agrico Canada Ltd fertilizer (available in large quantities at the site for hydrocarbon remediation in landfarms) prior to seeding. As the disturbed areas overlooked Exeter Bay, the use of this fertilizer was deemed inappropriate for this project since the MSDS literature indicated potential harm to aquatic organisms.

Salix arctica Pall. (willow) cuttings were harvested from a donor population near the beach after a frost in Sep- tember 2009. A shoot segment 30 - 150 cm long and ~1 - 2 cm in diameter was cut from the stem of each donor willow plant with pruning shears, and kept in water until time of transplantation (<10 hrs). Longer cuttings (>40 cm) were further separated into two or more cuttings before transplantation as recommended by Walter et al. [23] . Cuttings were folded in three, and then buried in holes (10 - 15 cm deep) that were created on an angle with a pick-axe (R. Lewis 2009, personal conversation). Care was taken to ensure that one or two leaf buds were above the surface of the soil. A total of 540 willow cuttings were planted in 26 plots of 12, 16, 20 or 25 cuttings(based on space and amount of collected material) to create small “islands” within the unvegetated area to accelerate growth of these slow-growing woody perennial species. Four to five cuttings were planted in a row, with four to five rows per plot. Cuttings were planted approximately 50 - 100 cm apart within a row, and the next row was offset by approximately 50 - 100 cm so cuttings in two adjacent rows were not directly in line creating a diamond-shaped pattern as recommended by Walter et al. [23] . The use of islands was chosen as a revegetation strategy as the areal extent of the unvegetated areas was too large to economically cover completely with cuttings [24] . Plots were randomly distributed across the disturbed area with cuttings located on flat and sloped areas.

Another 150 cuttings were transplanted in the fall of 2011. Cuttings were again collected following a frost from donor plants located in areas surrounding the disturbed area. Five plots were planted with 27 - 30 willow cuttings using the same method as those planted in 2009, whereas a sixth plot was planted with a modified approach in which willows were planted in long trenches in a staggered pattern so plants could lie in their natural orientation and nodal root growth could be encouraged [23] . These longer and shallower trenches were at times easier to dig in the rocky ground than vertical holes (as buried boulders need to be moved or planting sites repeatedly shifted). In addition to burying three quarters of the plant cutting in soil to encourage root development, the existing secondary root structures along the exposed region of the cutting were covered with small mounds of soil to promote nodal root development.

Twenty-two kg of Lolium multiflorum (annual ryegrass; Lot L19-8-22G) obtained from Country Depot in Kingston Ontario, was seeded by hand across ~7900 m² of the disturbed area at a rate of 29 kg/ha (commercial seeding rates = 15 - 30 kg/ha).

Ripe seeds and berries from native plants were collected by hand into five large ziplock bags (3.8 L each) from the areas above and below the disturbed area. Seeds were of different shapes, sizes, and weights based on their dispersal mechanisms. Each bag contained a different mix of seeds and berries based on the collection area; therefore seed density was not calculated prior to dispersal. Native seed was mixed by hand to create a more homogenous mix of seeds and berries and then dispersed by hand on the same area (~7900 m²) as the annual ryegrass. Seeds were not tested for germination prior to dispersal. Future studies are required to investigate seed germination and seedling survival for seeds from this area. Following seed dispersal (native and non-native), seeds were covered with soil by attaching a large metal fence section (~4 feet × 5 feet) to the back of an all-terrain vehicle, and dragging it perpendicular to the direction of the ridges to help push loose soil into the grooves to cover the seeds.

Plantassessments were conducted in late July and early August in 2010, 2011, and 2012 to determine the number of Salix arctica cuttings that survived the winter and produced new shoot growth. New shoot growth was qualified as the observation of one or more leaves, and quantified in 2010 as, i) limited growth (1-2 leaves), ii) moderate growth (3 - 4 leaves) and iii) significant growth (4+ leaves). In 2011 and 2012, cuttings were determined to be alive (one or more leaves) or dead (no leaves). Four cuttings were excavated in 2010 to observe root growth from one plant with limited growth, one plant with moderate growth, and two plants with significant growth.

Based on slow growth rates of Arctic plants, where most species are still relatively small during the first growing seasons [25] , the use of >20% frequency rather than >2% ground cover or >10 stems/m² was chosen to measure initial success for the establishment of seedlings on the remediated areas at DYE-M in 2010, 2011 and 2012 [7] [26] [27] . Twenty-eight transects and 136 1-m² quadrats were systematically established across the disturbed area in 2010to identify native plant species frequency and biodiversity. A more detailed assessment was conducted in 2011 using systematic transects and 379 1-m² quadrats [28] . Due to time and weather constraints, only 15 1-m² quadrats were randomly established for monitoring in 2012. Loliummultiflorum was only assessed in 2010 as it is an annual species. Lolium multiflorum frequency and density were assessed in 2010 by counting individual plants in each quadrat.

Prior to revegetation efforts at DYE-M, all fifteen species observed in the initial study were investigated to determine their revegetation potential based on: i) their ability to tolerate sandy soil, ii) their potential to improve the growth substrate, iii) whether they were perennials, iv) whether they were native to the Cape Dyer region, v) whether they could be obtained commercially or locally in sufficient quantities, vi) whether they had previously been shown to be successful in revegetation studies, and vii) whether they had the ability to prevent further erosion on the site (adapted from [7] ). Of the fifteen identified species, all were perennial and capable of reproducing by seeds or achenes, and some could also reproduce by rhizomes, stolons and rooting from prostrate shoots (Supplementary Information Table 1). Based on recommendations by R. Lewis from Arctic Alpine Seed Ltd (personal conversation, 2009) and the criteria above, it was determined that seeds from all species observed at Cape Dyer would be collected from the surrounding areas for assisted seed dispersal as commercial sources of native arctic seeds were not available in sufficient quantities. In addition, Lolium multiflorum(annual ryegrass) was recommended as a nurse species to help prevent erosion and to improve the nutrient content in the soil.

Ninety-four percent of the 540 S. arctica cuttings planted in 2009 survived the winter and produced new shoot growth in 2010. First year survival rates greater than 80% are considered successful for Salix sp. cuttings [32] , while Densmore et al. [33] reported survival rates between 86% - 96% for different lengths of Salix alaxensis cuttings after one year of growth in Alaska. In this study, S. arctica cuttings continued to develop new shoot growth after initial observations in mid-July. Catkins were observed on some plants during the survey period.

Root growth assessments determined that roots ranged in length from non-existent (plant with limited shoot growth), to extensive (plant with significant shoot growth) with roots >10 cm long and secondary and tertiary root development. Maximum root growth was observed in damp but not flooded areas. Visual observations of S. arctica from other areas around the site support the observation that plant growth is greatest in areas that offer refuge from strong winds (along rocks and boulders), as well as providing a significant source of soil water content (near water pools and along water run-off areas). Furthermore, these S. arctica tend to have well-developed secondary root growth along the full length of the stem (Figure 4).

In 2011, two landslides, one in June and one in late August, carrying rock and sediment debris from the road and hillside above the beach impacted six of the revegetation plots. Only thirty-five of the 88 willow cuttings from these plots survived. Initial observations from late July indicated that most cuttings in the other areas were producing leaves, but that secondary root development and shoot length growth was limited. This is to be expected as growth in the arctic is slow and significant change in size may take years of observation. Plant surveys were conducted six weeks later in September, and it was determined that some cuttings that had been judged unsuccessful in July had multiple leaves and appeared to be in significantly better health. Based on the September observations, 84% of the 428 S. arctica cuttings found in 2011 had survived since 2009. Results from 2010 and 2011 indicate that end-of-season observations provide a more realistic indication of overall plant health and survivorship than early or mid-season observations.

Plant surveys were conducted in August 2012, following another landslide in July which impacted a similar area as in 2011. Another two plots were compromised by this third landslide, leaving only 29 of the 128 willows in this area. In addition, road work destroyed another two plots (24 cuttings). Of the remaining 388 cuttings not impacted by landslides or other disturbances, 82% survived to year three. This three year survival rate of 82% is similar to observations from a revegetation effort following an oil spill along the Trans-Alaska pipeline in 2001, where 80% - 88% of transplanted willow cuttings survived after three years [34] . Observations of surviving plants at DYE-M indicated that many plants were starting to produce new stem tissue as well as new leaves, with many plants having 50+ leaves by the start of August, 2012. However, Densmore et al. [33] reported that long-term survival dropped to 53% after nine years, indicating that long-term monitoring and further transplantation of cuttings may be necessary to ensure long-term revegetation success. Of the 178 new cuttings planted in 2011, 93% survived the first winter and were producing leaves in 2012, indicating similar one year success as the first set of cuttings. The sixth plot planted with the alternative method had the lowest survival rate (89%), but the surviving plants had the most leaves along the aboveground shoot tissue. Encouragement of nodal adventitious roots has been shown to significantly increase stem length in Cucurbita pepo L. [35] . Methods to optimize root development and formation along stem tissues require further investigation in shrub species.

Lolium multiflorum (annual ryegrass) seeded at a planted density of ~29 kg/ha in 2009, was observed in 85% of

all quadrats, with a maximum density of 198 plants/100 cm². While the majority of L. multiflorum germinated and grew in 2010, a minority of seeds germinated in 2011 and 2012. No plants had developed spikes or inflorescences by the end of July, 2010, and all plants appeared so be stunted by the cold climate, with heights between 5 - 10 cm, in comparison to normal heights up to 125 cm [36] . The majority of plants were observed in furrows or around the edges of boulders where they were slightly protected from the wind. Dead shoot tissue from plants that grew in 2010 was no longer present in 2012. This material is expected to help introduce organic material into the top layer of soil. Thus while seeding densities of L. multiflorum between 3.5 to 30 kg/ha have been used in previous studies [8] [37] [38] , a higher planting density may be necessary in extremely exposed Arctic environments when used for erosion prevention and/or as nurse grass for native species to ensure growth of a sufficient quantity of shoot material. However, care must be taken to ensure that the selected species is not seeded too densely or it might prevent germination and growth of native species, particularly if it is a perennial species [16] [39] .

Previous studies have used various guidelines to determine the success of revegetation projects, including colonization by native plants and establishment of seedlings, increased species diversity, reproduction by colonizing species, development of a moss layer, accumulation of ground litter, and 85% total plant cover [40] -[43] . Quantitative criteria used by Kershaw and Kershaw [27] , and Rausch and Kershaw [7] to evaluate successful colonizers of long-term natural revegetation projects under extreme growth conditions included species that met at least one of, >2% ground cover, >20% frequency, or >10 stems/m². Seedlings and plants from 14 species were observed in the quadrats or along transects between 2010 and 2012, in comparison to the 15 species recorded from plant surveys of the surrounding areas in 2009. Ground cover was not determined on an individual or total species basis during the surveys due to the low species diversity per plot and small size of plants. Of the 14 species observed during these three years, Chamerion latifolium (L.) Holub was the only species to have greater than 20% frequency every year, including in 2009 (22% - 67%), (Figure 3). Other species with greater than 20% frequency during at least one year of monitoring included Cerastium uniflorum Clairv. (11% - 47%), mixed grasses (14% - 100%), Montia fontana L. (0% - 33%), Papaver radicatum Rottb. (2% - 73%), S. arctica (15% - 75%), and Silene acaulis (L.) Jacq. (7% - 42%) (Figure 3).

New species observed on the disturbed areas included Cochlearia officinalis L., Melandrium affine (J. Vahl ex Fr.) J. Vahl, and M. fontana [21] , while slow-growing heath-shrub species Cassiope tetragona (L.) D. Don, Dryas octopetala L., Empetrum nigrum, and Vaccinium uliginosum were not recorded on the remediated areas. In temperate locations, plant communities develop over time through a series of natural succession events, whereby initial colonizing species are gradually replaced by other species, often with slower dispersion techniques [44] . All 14 species observed in the remediated areas have wind or bird-dispersed seeds or produce bulbils, helping them to colonize disturbed areas in the first phase of plant succession. In natural environments following disturbance such as fire, the soil seed bank can accelerate site recovery, as seed banks generally represent the established vegetation in the area with the exception of woody perennial species [45] . However, this area has a history of repetitive disturbances, and the minimal topsoil that may have been present on the disturbed areas was not salvaged prior to landfill excavation. In future, preservation of this top layer of soil could introduce seeds and organic matter which can accelerate development of a natural plant community and provide the most widespread initial coverage, leading to long-term soil stability. This is particularly important in the Eastern Arctic, where no commercial seed mixture is currently available.

Despite the harsh climate and unpredictable weather patterns in the arctic that make revegetation work difficult, visual observations of the disturbed areas in 2012 indicated that patches of new vegetation are becoming more predominant compared to the almost complete lack of vegetation in 2009. Arctic plant communities are slow growing and seedling distribution tends to be skewed towards those sites that offer favourable conditions and are in close proximity to native tundra. Germination sites or microsites may be created naturally such as by desiccation cracks in the ground [46] , or may be created with assistance. For example, Cater et al. [37] used a track vehicle to scarify the ground prior to seeding. These microsites are important for trapping seeds during dispersal, and also create a microclimate with specific properties favoured by plants including a greater mean soil temperature, lower wind speed, increased soil moisture and superior soil nitrate levels [46] . In this study, most seedlings were initially observed growing within furrows or around windbreaks such as the edges of rocks, as seeds were less likely to be blown or washed away, and/or sheltered within depressions, which provide suitable microclimates for germination and growth. However, the furrows created in 2009 had almost completely been smoothed over by 2012. A higher density of new seedlings was also observed around S. arctica patches, indicating that S. arctica “islands” are likely trapping seeds blowing across the unvegetated areas, and providing safe sites and microclimates that enhance seed germination and seedling growth. In addition, shrub species such as S. arctica may protect and insulate herbaceous species during the winter from extreme temperatures and winds by trapping snow in the branches, which promotes plant establishment [47] . Thus initial observations indicate that loosening soil prior to seeding and creating S. arctica islands enhances revegetation and may increase the probability of long term success.