Reintroducing Vascular and Non-Vascular Plants to Disturbed Arctic Sites: Investigating Turfs and Turf Fragments

ABSTRACT

Disturbed low-Arctic environments provide many challenges for ecological restoration, from harsh climates and remote locations to limited knowledge on plant establishment and successional pathways within tundra ecosystems. Due to limited commercially available materials for restoration of native low-Arctic plant communities, transplantation may provide an effective technique for revegetation in these difficult-to-restore environments. In this study, whole-turfs and shredded turfs were harvested from undisturbed upland-heath tundra near Rankin Inlet, Canada, and transplanted onto nearby disturbed gravel quarries to investigate species survivability and development of upland-heath vegetative communities. Two years following transplantation, turfs were found to maintain 85% of the initial vegetative cover and 91% of the initial species richness, with expansion up to 8 cm into the surrounding substrate, and production of seeds and spores. Although shredded turfs were unable to significantly establish vascular species, evidence suggests a shredded turf may establish non-vascular plant cover over a larger area than intact turfs, if given greater protection from environmental stressors. Our results demonstrate that whole-turfs are resistant to harvesting and transplantation stresses, flooding, drought, and poor soil conditions, and are an effective means of species transfer promoting development of vegetative cover on disturbed substrates. High species survivability indicates that turfs have the potential to provide disturbed areas with a wide array of native species, critical for the development of sustainable and self-organizing assemblages of native vegetation.

Sensitive Arctic environments are experiencing increasing disturbance from resource exploitation (Forbes et al. 2001). The most frequent disturbances resulting from mining activity include gravel quarries, raised pads, and roads, which are known to result in severe disturbances including removal of vegetative and organic layers, increased soil pH, and altered hydrology (Naeth and Wilkinson 2014, Miller et al. 2021). Natural regeneration of vegetative communities within disturbed low-Arctic environments can be a slow process. Decades or even hundreds of years are required to fully establish native vegetative communities (Forbes and Jefferies 1999, Hodkinson et al. 2003), necessitating active restoration to recreate pre-disturbance or near pre-disturbance conditions. Low-Arctic environments provide considerable challenges for restoration practitioners. Extreme abiotic conditions, such as low temperature and precipitation and short growing seasons, can impede key ecosystem functions such as decomposition, nutrient cycling, and vegetative production (Naeth and Wilkinson 2014, Kearns et al. 2015, Mehlhoop et al. 2018).

The limited number of colonizing species (Forbes and Jefferies 1999), and lack of commercially available tundra plant propagules often leads restoration practitioners to use non-native species and/or fertilizer amendments to quickly establish cover (Hagen et al. 2014). These practices may alter the natural establishment of native species, leading to the promotion and persistence of non-native vegetation (Younkin and Martens 1987, Barni et al. 2007, Hagen et al. 2014, Kearns et al. 2015, Rydgren et al. 2016). Effective Arctic restoration is further challenged by limited knowledge on successional pathways, facilitative or inhibitive effects of common species, and the development

Restoration Recap

• • Whole-turf transplants demonstrate resilience to environmental stressors, maintain high degrees of vegetative cover and species richness, and demonstrate vegetative reproduction two years post-transplantation, allowing for low-cost re-introduction of native species onto disturbed substrates.

• • Shredded material may represent an effective transfer of non-vascular species but must be protected from wind, flooding, and desiccation. Selective harvesting depths, and greater quantities of shredded material may improve development of early successional non-vascular communities.

• • Short-term vegetative expansion from transplanted turfs is limited to 8 cm from the turfs, although further studies are required to determine if expansion is primarily above or below the soil surface.

• • Recovery of turf harvesting locations was minimal after two years, indicating that when possible whole-turfs should be harvested from areas prior to disturbance.

of complex plant-fungal or bacterial associations. Lastly, limited infrastructure necessitates the need for simple, cost-effective methods in these remote locations.

A common restoration method is whole-turf or sod transplantation (i.e., harvesting of intact vegetation and underlying soil layers and subsequent transplantation into disturbed sites). Turf transplants have been conducted in several Arctic (Kidd et al. 2006, Cater et al. 2015) and alpine environments (Bay and Ebersole 2006, Aradottir and Oskarsdottir 2013, Mehlhoop et al. 2018), however, the efficacy of this technique has not yet been examined in low-Arctic vegetative communities including upland-heath tundra. Documented benefits of turf transplantation include: the use of native, functioning assemblages of vegetation (Cater et al. 2015), inclusion of soil resources such as organic matter, soil invertebrates, microbes, and root systems (Conlin and Ebersole 2001, Klimeš et al. 2010), maintenance of species diversity (Aradottir 2012, Aradottir and Oskarsdottir 2013), and reduced shock to individuals within the transplants (Bay and Ebersole 2006). Additionally, turfs provide protection for germination of seeds and spores (Mehlhoop et al. 2018), vegetative propagation within and adjacent to the turfs (Klimeš et al. 2010, Hnatowich et al. 2022), and are simple to harvest and transplant.

Sod dumping or the spreading of shredded vegetation and soils is a technique frequently used to restore non-vascular communities within peatland environments (Rochefort et al. 2003, Rochefort and Lode 2006, Gauthier et al. 2018, Xu et al. 2022), and to transfer species within European meadows (Kiehl et al. 2010). Previous research has demonstrated thin layers of shredded tundra material can propagate non-vascular communities in northern environments (Aradottir 2012), suggesting that comparatively less material than required by turfs may be effective at initiating bryophyte establishment. Similar to turf transplantation, there have been very few studies on the use of sod dumping, or shredded vegetation within low-Arctic environments.

Here we investigated the use of both turf transplants and shredded tundra materials within two disturbed upland-heath tundra sites. Our first objective was to determine the efficacy of these two restoration methods in maintaining vegetative cover, community composition and species richness. We hypothesized that transplanted turfs would best maintain both vegetative cover and species richness. Furthermore, we hypothesized that the community composition of both turfs and shredded layers would change following transplantation, with turfs favoring the development of graminoids, and shredded layers promoting non-vascular colonization. Our second objective was to examine soil nutrient conditions in transplanted turfs. We hypothesized that nutrient concentrations would not change significantly two years post-transplantation. Our third objective was to investigate the recovery of turf harvesting locations. We hypothesized that compared with other tundra vegetation, moss species would have the greatest vegetative cover within the harvesting plots two years post-harvest.

Methods

Study Location

Our study was located at the Agnico Eagle Mines Ltd. Meliadine gold mine (63°01'22.9" N 92°11'41.1" W DMS) located near Rankin Inlet, Nunavut, Canada (Supplemental Material, Figure S1). Situated in the Southern Arctic Ecozone, the area is characterized by cold winters (30-year climate normal, seasonal average temperature^–24°C, November–April) and short, moderate summers (seasonal average temperature 7°C, June–September) with approximately 85 growing season days (Ecological Stratification Working Group 1995, Environment Canada 2022). Precipitation is low with an annual average of 310 mm, roughly half falling during the summer months, and annual average wind speeds of 23 km/h. During the study, the site experienced both high summer rainfall in 2019, with 140 mm during June–July and 192.2 mm during August–September, a dry early summer in 2020, with only 12.4 mm falling during June–July, and 99.3 mm during August–September, and in 2021, 87.1 mm during June–July, and 124.5 mm during August–September (Environment Canada 2022).

Soils are predominately Turbic Cryosols with Regosolic features, Turbic Cryosols, and Organics, with cryoturbation resulting in well-developed hummock-hollow complexes across the study site (Golder Associates 2014). Mineral soils in the local area are composed of well-drained sand, silt, and gravel tills usually overlain by thin (10–12 cm) organic layers. Heath tundra, heath-lichen and lichen-rock communities associated with well-drained soils represent over half of the vegetative communities within the local area. Common species include the shrubs Cassiope tetragona (Arctic mountain heather), Dryas integrifolia (white mountain-avens), Salix reticulata (net-leaved willow), and Vaccinium uliginosum (bog blueberry), forbs Oxytropis maydelliana (Maydell's oxytrope), Astragalus alpinus (alpine milkvetch), Cardamine digitata (Richardson's bittercress), and Stellaria longipes (long-stalked starwort), the mosses Aulacomnium turgidum (swollen thread moss), Pohlia nutans (nodding thread moss) and Hylocomium splendens (stairstep moss), and lichens Dactylina arctica (finger lichen), Thamnolia vermicularis (white worm lichen), Cetraria spp., and Cladonia spp.

Experimental Design

Several gravel quarries are present on the Meliadine mine site, two gravel quarries (Q1 and Q2), were selected for the restoration sites (Supplemental Material, Figure S1). The quarries were selected following the identification of areas not expected to have further anthropogenic disturbance, areas without significant dust from nearby roads, and flat, open areas with similar gravel substrates (Supplemental Material, Table S1). Upland-heath donor sites adjacent to the quarries were selected based upon similar community composition of both vascular and non-vascular vegetation, ease of harvesting/transport, and depth of both organic (~10 cm) and mineral (~5 cm) soil layers.

At each quarry restoration site, a 15 m2 area was delineated and a backhoe-loader was used to excavate quarry substrates and place four, 15 m long × ~50 cm high rows, spaced ~1.5 m apart (Supplemental Material, Figure S2). The rows were re-contoured by hand to simulate the hummock-hollow microtopography characteristic of the surrounding tundra landscape. Hummocks were ~50 cm high, with the hummock ridges spaced one meter apart. Each row contained 10 treatment plots (0.5 × 1 m) separated by 0.75 m (Supplemental Material, Figure S2). Each treatment plot contained one of four treatments, specifically either: 1) a single 40 cm × 40 cm turf placed in the hollow (T); 2) shredded turfs which were spread over the entire plot area (S); 3) a combination of turfs and shredded material (TS); or 4) a control plot with no material added (C) (Supplemental Material, Table S2). Treatments were placed in a randomized complete block design. Blocks were organized across the four rows at each site, for a total of ten blocks per site.

Harvesting of turfs and shredded material was conducted in upland-heath communities located near each restoration site (119 m and 60 m distance for Q1 and Q2, respectively). A flat-head shovel was used to cut, lift and remove the turfs from the harvesting sites. Twenty individual turfs were harvested at each harvesting site, with at least one meter of separation between each harvested turf. The underlying material was predominantly organic matter; however, effort was taken to ensure that harvested material also contained some of the underlying mineral layers (Supplemental Material, Table S1). Harvesting plots were flagged, spatially referenced, and photographed. At all harvesting plots, the depth of each directional face was recorded for monitoring future vegetative encroachment or ground subsidence.

All turfs (n = 40) were placed with the vegetative surface level flush with the surrounding substrate surface. Once placed, the surrounding substrate was pushed against the turfs to ensure good turf-substrate contact. Three additional turfs were harvested for shredding, which provided enough shredded material to cover ~30% of a treatment plot with 2 cm of shredded material. Shredded material was sieved through a 2 cm × 2 cm metal mesh screen to separate organic layers, and shred vascular and non-vascular plants into fragments ~1–3 cm in size, which were then homogenized before application. Any leftover shredded material was discarded. All treatments were covered by a jute-mesh erosion control blanket (Anti-wash GEOJUTE®, Belton Industries, Honea Path, SC) that was pinned in place using metal staples. The jute-mesh was selected over other cover materials, such as straw, due to the remote location and high wind speeds that could easily blow away cover materials. The jute-mesh had high light penetration (45%–60%, mesh size of 1.3 × 1.3 cm to 1.9 × 1.9 cm), ability to protect from erosive forces (wind/water), water storage capacity (425%) and degradation time of approximately 1–2 years, depending on climate.

Vegetation and Substrate Analysis

Immediately following transplanting (July 7th–12th, 2019), a 0.16 m2 gridded quadrat (25, 8 cm × 8 cm sub-quadrats) was placed over the top of each T, TS, and S treatment. Values ranging from one to four (1 = < 25%, 2 = 25–50%, 3 = 50–75%, 4 = > 75%) were used for estimation of species cover in each individual grid. All vascular and non-vascular plants within harvested turfs were identified to species level whenever possible. Absolute and relative cover were calculated as follows: absolute percent differences = ([species cover in 2021 – species cover in 2019)/species cover in 2019], and relative percent differences = ([species cover in 2021/total species cover in 2021] – [species cover in 2019/total species cover in 2019]). For T treatments only the center plot was surveyed. For TS and S treatments the center plot and the inside of both sides of the hummock were surveyed in the same fashion. A rapid visual survey was completed to ensure no vegetative cover in the control plots. All treatments were resurveyed between July 7th–13th, 2021 in a similar manner, except for T treatments where the inside of both sides of the hummocks were included. Aboveground expansion was assessed in each treatment by surveying an additional 24 cm on the non-treated sides of the central plot, to avoid inclusions of vegetative material within shredded layers of S and TS treatments. The same 0.16 m2 gridded quadrat was used for surveying expansion. Turf harvesting locations were surveyed between July 15th–17th, 2021 using the same method, except the 0.16 m2 gridded quadrat had four 20 × 20 cm sub-quadrats.

Soil Sampling and Analysis

In July 2019 soil samples were collected within 1 m of ten randomly chosen turf harvesting locations at each restoration site. Both organic and mineral horizons were sampled. A composite sample of the substrate used to create the hummock-hollows at Q1 and Q2 was also sampled. The same harvesting locations were sampled between July 15th–17th, 2021. In addition, on July 16th–17th, 2021 composite soil core samples were taken (n = 5 cored samples per turf, PN009 dry sampling tube, JMC, Iowa, USA) from all treatments that included a turf (n = 40). Organic and mineral layers were sampled simultaneously, then separated into their respective layers, with the soil corer wiped clean between each sample. All samples were transported to the University of Saskatchewan where they were air-dried at room temperature for roughly seven days, sieved (4 mm2), and stored at–20°C for future analysis. Water extractions were conducted on all soil samples using a 1:4 ratio of soil to Milli-Q water, except for eight (2019) and 23 (2021) samples that required higher water to soil ratios (i.e., 1:6 or 1:8) due to high amounts of organic matter. Extracts were then measured on a Dionex ICS-2000 Ion Chromatograph to determine concentrations of cations (Ca²⁺, Mg²⁺, K⁺, Na⁺, NH₄⁺) and anions (F^–, NO₃^–, NO₂^–, SO₄^–, Cl^–, PO₄3^–). pH was measured using a Mettler Toledo FiveEasy pH meter (Columbus, OH). Subsamples of each soil were ground in preparation for total carbon (TC), total inorganic carbon (TIC), and total nitrogen (TN). Total nitrogen was analyzed using a LECO TruMac CNS Analyzer (Midland, ON, Canada), total inorganic nitrogen (TIN) was calculated from concentrations of NO3^–and NH₄⁺, and total organic nitrogen (TON) was calculated by the difference between TN and TIN. TIC was analyzed by an acetic acid pH standard curve (20.2, CSSS, 2008), TC was analyzed with a modified high-temperature combustion method (21.2, CSSS, 2008), and total organic carbon (TOC) was calculated by the difference between TC and TIC. TC and TIC were analyzed by ALS Environmental (Saskatoon, Saskatchewan, Canada).

Statistical Analysis

Statistical analysis was conducted in R 4.1.2 (R Core Team, 2020, Vienna, Austria). Vegetative percent cover data used in multivariate analyses were first converted into a Bray-Curtis distance matrix. Community composition of all treatments during 2021, and community composition of turfs (turfs of T and TS treatments) in 2019 and 2021 were visualized via non-metric dimensional scaling (NMDS) ordination plots using the 'nmds' function within the 'vegan' package (Oksanen et al. 2020). Differences in community composition between treatments in 2021 (with block nested within site), and between turf community composition in 2019 and 2021 (with each turf given a unique identifier "subject ID" to account for repeated measures) were tested using permutational multivariate analysis of variation (Per-MANOVA) using the 'adonis' function in vegan. Post-hoc comparisons between treatments were determined using 'TukeyHSD' in the 'stats' package (R Core Team 2020), on the centroid location of each treatment.

The effects of treatment over the entire treatment plot (hummocks and hollow) on the response variables: species richness, cover of bare ground (including rocks and stones), litter, total vegetation, growth forms (deciduous and evergreen shrubs, forbs, graminoids, biological soil crusts, lichens, and mosses), and presence/absence of flowers (i.e., number of quadrat grids containing flowers, per species) in 2021 were assessed through linear mixed effect (LME) models ("lmer" functions within the "lmerTest" package [Kuznetsova et al. 2017]) and one-way analysis of variance (ANOVA) ("anova" within the "stats" package), using block as a random factor (each block at each site was given a unique identifier). All data was inspected to ensure assumptions of ANOVA tests were met. When data did not meet assumptions, data were transformed (detailed transformation information can be found in Supplemental Materials, Tables S4, S6, S8, S11, S13, S14, S15). Changes in the above response variables were also examined within shredded materials (S and shredded layers of TS treatments) between 2019 and 2021 and within transplanted turfs (turfs of T and TS treatments) between 2019 and 2021 using the same approach. Differences between years, sites, and their interaction were tested with LME models and two-way ANOVA (with subject ID as a random factor). Expansion of vascular and non-vascular cover between treatments and distances (0–8, 8–16, 16–24 cm) from the treatment plot were assessed with LME models and two-way ANOVA tests, using block as a random factor. Expansion from turf treatments (bare ground, litter, vascular, and non-vascular cover) between distances (0–8, 8–16, 16–24 cm) were also assessed with linear models and one-way ANOVA.

Mineral and organic soil layer pH, cations, anions, TON, TIN, TOC, and TIC were compared between transplanted turfs and their harvesting locations in 2019 and 2021

Figure 1.

Non-metric dimensional scaling (NMDS) ordination plots of treatment community composition two years after turf transplantation. Treatments are represented by circles (control), squares (Turf), triangles (Turfs + Shredded), and diamonds (Shredded). Ellipses represent the standard error of the weighted average of scores.

for each site separately using linear models and one-way ANOVA tests. All data was inspected to ensure assumptions of ANOVA tests were met. When assumptions were not met, data were transformed (detailed transformation info can be found in Supplemental Material, Table S18). Post hoc comparisons using the 'lsmeans' function within the 'lsmeans' package (Lenth 2016) were used to determine statistical differences between groups in all linear models and ANOVA tests.

Results

Comparison of Treatment Plant Communities

Two years following turf transplantation, significant differences in community composition between the four treatments were observed (PerMANOVA, df = 79, R² = 0.43, p = 0.001) (Figure 1) (Supplemental Material, Table S3). The community composition of all transplanted treatments were significantly different from the control (Tukey's, p = > 0.001), with S treatments also significantly different from T and TS treatments (Tukey's, p = 0.009). T and TS treatments did not differ (Tukey's, p = 0.991). T and TS treatments contained the highest species richness, greatest vegetative cover, and greatest presence of flowers, whereas S treatments contained approximately half of the species richness, approximately one-quarter of the vegetative cover, and approximately one-tenth the flowers of T and TS treatments (Table 1, Supplemental Material, Tables S4–6). All growth forms, except for biological soil crust and lichen cover, were significantly higher in T and TS treatments compared to S or C treatments (Table 1, Supplemental Material, Table S4).

Post-transplant Changes in Community Composition of Shredded Materials

Shredded layers (i.e., S and shredded material from TS treatments) significantly decreased in species richness post-transplantation (ANOVA, F_1,39 = 9.91, p = 0.003) (Supplemental Material, Tables S7–8). Cover of total vegetation (F_1,39 = 72.68, p = < 0.001), bryophytes (F_1,39 = 57.58, p = < 0.001), and lichens (F_1,39 = 70.92, p = < 0.001), decreased by at least 50%, while evergreen (F_1,39 = 167.69, p = < 0.001) and deciduous shrub cover decreased at least 80% (F_1,39 = 86.29, p = < 0.001) (Supplemental Material, Tables S7–8). Both total litter (F_1,39 = 226.96, p = < 0.001) and bare ground (F_1,39 = 167.1, p = < 0.001) increased in cover over the two years. The only growth form to increase were forbs (F_1,39 = 64.60, p = < 0.001), which demonstrated an 18-fold increase in cover, likely due to the colonization of volunteer species (e.g., Descurainia sophioides [northern tansy mustard], S. longipes, and several unidentified juvenile forbs). No evidence of biological soil crust development was observed within shredded layers.

Post-transplant Changes in Community Composition of Turfs

Turfs (i.e., T and central plot of TS treatments) maintained their species richness over two years (Supplemental Material, Tables S10–11). Although overall richness remained unchanged, plant community composition within the turfs changed significantly between 2019 and 2021 (PerMANOVA, df = 79, R² = 0.16, p = 0.001) (Figure 2) (Supplemental Material, Table S9). Significant differences were the result of increased cover of bare ground (ANOVA, F_1,39 = 27.87, p = < 0.001), litter (F_1,39 = 248.6, p = < 0.001),

Table 1.

Means, standard deviations and significant difference indicators for all treatments, two years following transplantation, on species richness, and cover of bare ground features, total vegetation, litter, and growth forms. Different letters indicate significant differences between treatments for each variable determined through post-hoc comparisons of linear mixed effect models using block as a random factor.

and forbs (F_1,39 = 14.90, p = < 0.001) along with decreases in total vegetation (F_1,39 = 13.19, p = < 0.001), evergreen shrub (F_1,39 = 31.04, p = < 0.001), lichen (F_1,39 = 32.96, p = < 0.001), and moss cover (F_1,39 = 7.04, p = 0.012) (Supplemental Material, Table S10–11). Turfs between the two sites were similar in composition, except that Q1 had higher lichen cover in both 2019 and 2021 (F_1,38 = 11.24, p = 0.002), whereas Q2 had higher cover of deciduous shrubs in both years (F_1,38 = 22.78, p = < 0.001) and higher cover of mosses, although only significantly in 2021 (F_1,38 = 11.13, p = 0.002) (Supplemental Material, Tables S12–13).

Evergreen shrubs demonstrated the largest decrease in absolute cover over two years (–31%), and while all species declined in cover, C. tetragona, Rhododendron lapponicum (lapland rosebay), and Rhododendron tomentosum (northern Labrador tea) declined more than other species such as D. integrifolia and Vaccinium vitis-idaea (lingonberry) (Table 2). Lichens had the second-largest decline in absolute cover (^–30%), with all species including T. vermicularis, Alectoria ochroleuca (green witch's hair lichen), D. arctica, Cetraria spp., and Cladonia spp. declining by over 25% in absolute cover. Some moss species declined in cover (e.g., H. splendens and P. nutans), while others increased (A. turgidum and Bucklandiella microcarpa [small-fruited rock moss]). Forbs and graminoids were the only growth forms to increase in cover (83% and 63%, respectively), notably from increases of Hedysarum alpinum (alpine sweetvetch), Bistorta vivipara (alpine bistort), C. digitata, O. maydelliana, and Carex bigelowii (Bigelow's sedge) (Table 2).

Post-transplant Vegetative Expansion

T and TS treatments had a significant expansion of vascular plants (F3,209 = 20.79, p = < 0.001) within 8 cm of the turf, roughly 8-fold greater than observed in S or C treatments Similarly, T and TS treatments demonstrated significantly higher non-vascular cover than C treatments (roughly 2- to 3-fold) (F3,209 = 10.60, p = < 0.001) with the cover of both metrics significantly decreasing after 8 cm (F_2,117 = 32.36, p = < 0.001; F_2,117 = 8.41, p = < 0.001) (Supplemental Material, Tables S14–15). Natural recolonization of C treatments resulted in roughly 2% cover of both vascular and non-vascular components within 8 cm of the treatment plot (Figure 3). Across both T and TS treatments, vascular and non-vascular species represented approximately 17% and 6% of total cover within the first 8 cm, respectively, quickly dropping to 4% and 3% after 8 cm, with higher standard deviations indicating highly variable expansion, i.e., small pockets of vegetation, or individual plants emerging near the boundaries of the turf (Figure 3).

Within 8 cm of the turf, the relative contribution of deciduous and evergreen shrubs, forbs, and graminoids was nearly equal, with each growth form contributing 27%, 27%, 26%, and 20% to vascular cover. Vaccinium uligonosum and Empetrum nigrum (crowberry), made the highest contributions of 18% and 10% of the relative vascular cover immediately adjacent to the turfs, with all other species representing less than 7% (Figure 4). Beyond 8 cm, forbs and graminoids dominated the expanding vascular cover, with 46% and 40% relative cover, respectively. Stellaria longipes, juvenile unidentified forbs, C. bigelowii, and Poa arctica (arctic bluegrass) contributed the most to the cover of vascular species beyond 8 cm (Figure 4). Deciduous shrubs represented 11% of the vascular cover beyond 8 cm, with evergreen shrubs only 3%. A similar pattern was observed with non-vascular cover, with lichens dominating the cover within the first 8 cm, representing 63% of non-vascular cover, and mosses representing 36%. Alectoria ochroleuca, Cetraria nivalis (crinkled snow lichen), Gowardia nigricans (grey witch's hair lichen), T. vermicularis, and unidentified crustose lichen growth represented 13%, 20%, 10%, 13%, and 11% of non-vascular cover immediately adjacent to the turfs, respectively (Figure 4). Beyond 8 cm, the relative contribution to non-vascular cover switched with mosses representing 63%, and lichens 36%. Further from the turfs,

Table 2.

Percent change of growth forms and species that represented at least 0.5% relative cover across both sites, either in 2019 or 2021, within the center plot of T and TS treatments. The percent difference of absolute cover is presented first, followed by relative cover in parenthesis. Absolute and relative cover are calculated as follows: absolute percent differences = ([species cover in 2021 - species cover in 2019]/species cover in 2019), and rela- tive percent differences = ([species cover in 2021/total species cover in 2021] - [species cover in 2019/total species cover in 2019]). Asterisks denote significant differences between 2019 and 2021. Means and standard deviations of each growth form can be found in Supplemental Materials, Table S10.

juvenile mosses comprised nearly half of all non-vascular cover, although T. vermicularis and crustose lichen growths still contributed > 10% each.

Recovery of Turf Harvest Sites

After two years, vegetative establishment within harvest plots was minimal. Vegetative cover was dominated by non-vascular lichens and mosses that had likely blown into the depressions left from harvest, accounting for 29% and 11% of the relative vegetative cover respectively (Supplemental Material, Table S16). Vascular plant growth originated primarily from the sidewalls of the harvest plot (i.e., new growth from cut and exposed belowground structures) and was composed of evergreen shrubs, deciduous shrubs, forbs, and graminoids at 14%, 12%, 9%, and 6% relative cover respectively (Supplemental Material, Table S16). There was evidence of slumping from the sides of some harvest plots, with the average depth of harvest plots decreasing by roughly one centimeter over the two years since harvest (i.e., an average of 10.29 cm to 9.13 cm).

Figure 2.

Non-metric Dimensional Scaling (NMDS) ordination plot of community composition of turfs over two years. Squares represent turfs from T treatments and the central turf from TS treatments. Years are represented by grey (2019) or black (2021) shading. Ellipses represent the standard error of the weighted average of scores.

Figure 3.

Heatmaps of mean relative vascular and non-vascular cover expanding from the central plot of each treatment, grouped by distance from the central plot. Different letters within boxes represent significant differences between treatments and between distances, with mean and standard deviations presented below.

Post-transplant Changes in Soil Nutrients

Two years following transplantation, the turfs within T and TS center plots demonstrated significant differences in nutrient concentrations from their respective harvesting sites, in both organic and mineral layers. Within organic layers, both sites demonstrated significant increases, approximately 15-fold (Q1) and 3-fold (Q2), in concentration of Cl^– (F_2,37 = 17.80, p = < 0.001; F_2,37 = 11.21, p = < 0.001) and approximately 19-fold (Q1) and 5-fold (Q2) in Na⁺ (F_2,37 = 16.59, p = < 0.001; F_2,37 = 24.96, p = < 0.001). Q1 demonstrated significantly decreased organic nitrogen (F_2,37 = 6.57, p = < 0.01), whereas Q2 demonstrated 9-fold increase in SO₄²⁺ (F_2,37 = 82.46, p = < 0.001), roughly twice the concentration of Ca²⁺(F_2,37 = 10.36, p = < 0.001), and significant decreases in NO₃^– (F_2,36 = 22.19, p = < 0.001) (Supplemental Material, Table S17).

Differences between transplanted turf and harvest sites within mineral soils were far more frequent than organic layers, with both sites demonstrating significant increases in pH (F_2,37 = 48.75, p = < 0.001; F_2,37 = 114.09, p = < 0.001), Cl^–(F_2,37 = 74.52, p = < 0.001; F_2,36 = 20.29, p = < 0.001), Na⁺ (F_2,37 = 40.77, p = < 0.001; F_2,36 = 73.31, p = < 0.001), Ca²⁺ (F_2,36 = 22.95, p = < 0.001; F_2,36 = 9.86, p = < 0.001), SO₄²⁺

Figure 4.

Heatmaps of vascular and non-vascular species expanding from the turfs that were observed in at least 10% of the T and TS treatment plots at either Quarry #1 or #2. Darker shades represent a greater relative contribution to the expanding communities. Bare ground, litter, vascular and non-vascular cover represent cover relative to all species, while deciduous shrubs, evergreen shrubs, graminoids, and forb growth forms represent cover relative to vascular cover, and lichen and moss growth forms represent cover relative to non-vascular cover. Different letters within the bare ground, litter, vascular, and non-vascular cover columns represent significant differences between distances.

(F_2,37 = 30.27 p = < 0.001; F_2,36 = 40.32, p = < 0.001) and inorganic carbon (F_2,12 = 23.92, p = < 0.001; F_2,12 = 23.47, p = < 0.001) (Supplemental Material, Table S17). Mineral soils of turfs at Q1 also increased in NO3^– (F_2,34 = 31.64, p = < 0.001), K⁺ (F_2,37 = 13.62, p = < 0.001), Mg²⁺ (F_2,36 = 17.93, p = < 0.001) and inorganic nitrogen (F_2,37 = 21.49, p = < 0.001) compared with harvest sites. Substrates used to create the hummock-hollow complexes had higher pH and greater concentrations of Cl^–, SO₄²⁺, NO3^–, Ca²⁺, and Na⁺ relative to harvest sites mineral soils (Supplemental Material, Table S17). Mineral soils at both sites demonstrated 2-fold and 10–fold increases in Ca²⁺, and SO4²⁺, respectively. Q1 mineral soils demonstrated 24-fold and 18-fold increases in Cl^– and Na⁺, respectively, compared to a 3-fold and 4-fold increase at Q2.

Discussion

We examined the revegetation of disturbed low-Arctic sites through transplanting upland-heath whole-turfs and shredded layers, finding whole-turf transplants were most effective at maintaining species diversity, community composition, and vegetative cover of all vascular growth forms. Shredded treatments performed poorly, while some development of non-vascular communities was evident, greater protection from external forces such as wind, water, and desiccation are required for shredded treatments to be effective. Changes in community composition occurred for both turfs and shredded treatments, and our hypothesis that turfs would favor graminoids, and shredded layers would favor non-vascular colonization was supported. Contrary to our hypothesis, nearly all metrics regarding available nutrients changed significantly within transplanted turfs; however, these changes were primarily driven by an influx of soluble salts from quarry substrates. We hypothesized that mosses would make the greatest contribution to vegetation cover in the early stages of turf harvest plot recovery; instead, lichens provided the greatest cover. Regardless, recovery of the turf harvesting plots were minimal two years post-transplantation, suggesting that the use of turf transplants should be pre-emptively planned to reduce additional disturbance in low-Arctic sites.

Intact versus Shredded Turfs Require Different Management

In 2021, turf transplants maintained high degrees of vegetative cover (85% of the average absolute vegetative cover recorded in 2019) and species richness (91% of average), along with nearly nine times the flower production of S and C treatments. In contrast, shredded materials only maintained 43% of the vegetative cover and 79% of the species richness of the original material placed in 2019. Unusually high precipitation during 2019 resulted in the flooding of several treatment plots. The flooding may have moved shredded material (evidenced by ~50% reduction in organic material within treatment plots) and the dry early summer of 2020 may have desiccated whatever remained within treatment plots, with further movement of shredded material caused by the high wind characteristic of the study site. Despite the use of an erosion control mat, shredded materials will require even greater protection to ensure optimal growth, whereas whole-turfs endured transplantation and extreme weather more successfully.

Post-transplant Changes in Shredded Community Composition

The establishment of non-vascular species from the shredded treatments is similar to previous research, i.e., shredded materials demonstrate non-vascular development (Aradottir 2012) and transfer of species from donor sites (Kiehl et al. 2010). Previous experiments have used up to 50 cm deep shredded layers, compared to the 2 cm depths of this study. The whole-turfs used to generate shredded material included the underlying organic and mineral soils, potentially diluting non-vascular propagules that are constrained to the upper ~5 cm of tundra mats. The thin layer of shredded material, limited non-vascular propagules, and sensitivity to desiccation (Kiehl et al. 2010) may have limited the effectiveness of our treatment. It must be noted that comparisons between shredded material and whole-turfs represent plot-by-plot comparisons and do not consider differences in the amount of material used for each treatment. One shredded turf provided enough material to cover the same surface area of five whole-turfs. Our study was not intended to address the differences between treatments regarding amount of material used, rather our study was an initial investigation into the efficacy of shredded and whole-turf treatments, and a determination if either treatment had merit in restoration of low-Arctic environments. Moreover, the significant establishment of non-vascular cover in shredded treatments warrants future comparisons between these treatments, perhaps through comparisons of biomass produced per area disturbed/soil volume, to further elucidate how these techniques may be used to restore large-scale disturbances.

Transfer of vascular species via shredded materials was very limited, indicating that physical shredding resulted in significant damage to vascular structures. Larger fragment size may promote the transfer of more intact vascular material and should be considered. Furthermore, the application of non-vascular rich shredded materials (i.e., harvesting only the top 5–10 cm of the tundra), spreading at greater depth, and enhanced protection from wind and water erosion may also improve establishment.

Post-transplant Changes in Turf Community Composition

Changes in the community composition of turfs post-transplantation confirmed our second hypothesis, and were similar to previous turf transplant studies within both Arctic (Kidd et al. 2006, Cater et al. 2015) and alpine environments (Bay and Ebersole 2006, Aradottir 2012). In addition, in a growth chamber experiment using upland-heath communities from the same study site, we observed increases in forbs and graminoids and decreased lichen cover (Hnatowich et al. 2022). Increases in graminoid growth forms in turfs are frequently reported (Bay and Ebersole 2006, Cole and Spildie 2006), and it has been theorized that graminoid expansion occurs to take advantage of increasing inorganic nitrogen from mineralization associated with harvesting disturbance (Aradottir and Oskarsdottir 2013).

Others have reported reductions in species adapted to low-nutrient conditions, such as evergreen shrubs, due to increasing nutrient availability (Zamin et al. 2014, Gu and Grogan 2020). Increasing pH may also have led to the significant decline in evergreen shrubs, and simultaneous increase in forb cover, with previous investigations finding evergreens, and associated fungal-dominated microbial communities are adapted to low-pH systems (Eskelinen et al. 2009). Previous turf transplants have also shown decreases in evergreen shrubs with turfs ≤ 30 cm² (Aradottir 2012), suggesting that evergreens generally do not weather separation from the surrounding tundra well, potentially damaging root systems and fungal associations. Natural colonization studies indicate late-stage species such as D. integrifolia usually establish after ~60 years, with C. tetragona establishing after ~100–150 years, and only with acidification of underlying substrate and significant build-up of organic material (Hodkinson et al. 2003). Therefore, late successional evergreen shrubs are likely not good candidates for restoration, even via intact turfs.

Post-transplant Changes in Soil Nutrients

Nearly all nutrients demonstrated significant differences from harvest sites, often demonstrating much higher concentrations, notably of Cl^–, SO42^–, NO3^–, Mg²⁺, Ca²⁺, and Na⁺, leading us to reject our third hypothesis. Initial investigations of metal concentrations within local soils demonstrated high concentrations of various ions (Golder Associates 2014), suggesting that the observed increases were likely due to the movement of these ions in soil solution across the quarries, especially during the heavy rainfall event in 2019. Overall, changes in the nutrient status of transplanted turfs were driven by nutrient conditions of the surrounding quarry substrate. In addition, changes in the mineral layer were likely influenced by the inclusion of some underlying substrate during turf soil sampling. The ability of whole-turfs to withstand an influx of soluble salts from the surrounding substrate suggests that turfs may support vegetative establishment even under poor soil conditions.

Vegetative Expansion

Expansion of vascular species from the treatment plots was largely confined to T and TS treatments and was minimal, with less than 25% of the immediate surroundings having any vegetative cover. Previous investigations of turf expansion have found 74% increases in turf area in Alaskan tundra (Shirazi et al. 1998) and 55% in Icelandic alpine heath (Aradottir 2012), over two years. Our expansion was less than reported by these studies (i.e., only ~17% vascular cover within 8 cm), however, a previous investigation into expansion using the same upland-heath communities found over 90% of expanding biomass was contained within belowground structures (Hnatowich et al. 2022). Early expansion from these transplanted upland health communities may be driven by belowground expansion, and further studies that incorporate both above and below-ground expansion are needed.

Vegetative expansion may have also been limited due to the conditions of the surrounding substrates, generally composed of coarse gravel, rocks, and stones with limited concentrations of soil organic matter. Previous investigations into vegetative establishment within Arctic and alpine environments have stressed both the importance of fine particles and soil organic matter to ensure establishment (Naeth and Wilkinson 2014, Mehlhoop et al. 2018). Furthermore, high concentrations of soluble salts originating from the substrates may have had a significant impact on expansion, suggesting that amelioration of surrounding substrates may be required to ensure turf survival and subsequent expansion.

Due to the observed changes in turf community composition, the observed expansion, and results from previous research, we recommend that restoration practitioners working in upland-heath environments focus on transplantation of forb and graminoid-dominated communities, particularly species from the Fabaceae family, notably A. alpinus, H. alpinum, and O. maydelliana for their survival rates, nitrogen-fixation capacities, and previously observed expansion. Carex species, C. digitata and B. vivipara demonstrated positive survival and growth, along with S. longipes, known as a highly polymorphic species and aggressive colonizer. A few select shrub species are also recommended, such as the deciduous S. reticulata, the only shrub with increased growth within the turfs, and the evergreen D. integrifolia, which had negligible decreases in cover. Both shrub species have shown high survival rates in previous research and were the predominate shrubs responsible for belowground expansion under growth chamber conditions (Hnatowich et al. 2022). Additionally, if a seed source could be obtained for the native annual D. sophioides, this may be an excellent candidate for quick establishment of vegetative cover, development of organic layers, and organic matter enrichment of the underlying soils over non-native species that carry risks of spreading to surrounding tundra.

Harvest Recovery

Harvest recovery was very limited across all harvesting locations, with the greatest vegetative cover represented by lichens that had likely blown into the harvesting plots and vascular ingrowth from the surrounding tundra. Mosses represented a minor component of vegetative cover within harvesting plots, rejecting our last hypothesis. Due to the limited recovery, we recommend that transplanted turfs not be harvested from undisturbed areas, rather, turfs should be harvested pro-actively from areas planned for future disturbance (Aradottir and Oskarsdottir 2013, Cater et al. 2015).

Conclusions

Overall, our two-year study demonstrated that turf transplants can maintain vegetative cover and species richness after two years, and can demonstrate resilience to flooding, drought, and an influx of less-than-desirable concentrations of soluble salts, potentially allowing their use in areas expected to be affected by extreme abiotic conditions, areas with poor substrate conditions (low organic matter and/or high concentration of soluble salts). Successful transplanting combined with the limited resources, time, and effort required to transplant turfs suggests this technique may be an effective means of re-establishing native vegetative communities onto disturbed sites within low-Arctic tundra, without the reliance on non-native species or fertilization. Importantly, while community composition of turfs may change and therefore are not identical to predisturbance conditions, most species transferred are likely to survive transplantation, providing disturbed areas with the necessary species required to regenerate self-organizing assemblages of native vegetation. Future monitoring will be important to conclusively determine the community alterations and species survivability past the two years analyzed in this study. Shredded turfs, while demonstrating establishment of non-vascular species over a wide area, demand greater protection from environmental conditions, limiting the sites within low-Arctic environments where this technique is effective. However, targeting non-vascular propagules for shredded treatments, increasing the depth of applied materials, the use of larger shredded fragments, and providing surface protection may increase the feasibility of this approach. Finally, due to limited recovery of turf harvesting locations, turfs used for transplantation should not be harvested from undisturbed sites, rather, plans should be made pro-actively to ensure that turf materials are harvested and stored from areas slated for development.

Supplementary Material

Click here to view supplementary material. (PDF)